A Giš on a Tree: Interactions between Images and Inscriptions on Neo-assyrian Monuments

The study of the relationships between images and writing in the Ancient Near East is a broad field, which includes different approaches of historical and philological analysis. In this article, I will focus on an aspect of such relationships that has been only partially analysed in previous studies, that is, visual and semantic interactions between images and inscriptions in some Neo-Assyrian monuments when part (i.e. one or more cuneiform signs) of an inscription interplays with the part of the sculpted image (i.e. one element of the image or part of it) that it crosses. 
The purpose of this article is to try to answer the following questions: is it possible to positively identify interactions of this kind? Is it possible to interpret their meaning and to understand their function (or functions)? Were they meant to be a sort of con- cealed “game” within the inscription, or were they meant to be seen? And once they were seen, who was the intended recipient, the audience, of such interactions? As will become evident in the course of the article, my belief is that there are no unequivocal answers to these questions.

from three bas-reliefs on orthostats attached to the walls of three different rooms of the palace: specifically, one on relief No. 23 from throne room B, one on No. 4 from room H, and two on No. 3 from room G.3 Besides being the first and one of the best-preserved cases in Assyrian architecture in which bas-reliefs were used to decorate almost all of its public spaces, the NWP is the only Assyrian building in which more than half of its reliefs are crossed in their central part by an inscription of ca. 20 lines. All of the inscriptions are more or less exact copies of the same text, repeated hundreds of times throughout the palace, the so-called Standard Inscription (SI).4 I will first try to analyse how to define an image-inscription interaction. When, in other words, can we almost certainly identify an intentional interaction (as opposed to a mere coincidence), and what are its main visual and, possibly, semantic characteristics. Namely, the de-contextualization of one or more signs of the inscription, the de-contextualization of the corresponding part of the image, the direct or indirect relation between the two of them, and the possible meaning and/or message inferred by the visual-literary "game" resulting from their interplay.
Secondly, I will present four examples of interactions coming from the NWP with some discussion of previous interpretations.
Then, I will try to show how the interactions on the NWP's orthostats should be considered as part of a broader compositional project that was behind at least some of the most important reliefs of the palace, a project that involved the SI and the ways in which it was engraved on the palace's slabs.
I will then try to analyse whether these interactions were meant to be visible to the public or not. I will argue that the nature of both Ancient Near East figurative 1956), by the Iraqi Directorate General of Antiquities  and by a joint expedition between the IDGA and a Polish mission directed by J. Meuszyński (1974-75), see Postgate/Reade 1980, 304-307, andOates/Oates 2001, with previous bibliography. For a study that combines NWP's reliefs and inscriptions, see Russell 1999. The reconstruction of the decorative program of the NWP, with the exact position of each of the surviving sculpted orthostats, was carried out first by J. Meuszyński (1981) and then by S. M. Paley and R. P. Sobolewsky (1987). See also Russell 1998. 3 The first two examples have already been noted by Brian Brown in his article on the NWP's reliefs and cultic architecture (Brown 2010), but while I agree with the majority of his conclusions, I believe that something can be added to his analysis. The interactions from slab G-3 were noted by me during a recent visit to the British Museum conducted for the purpose of this research. 4 RIMA 2 A.0.101.23. The SI contains all the topics of a typical royal inscription: the king's name and genealogy, some of his epithets, mentions of his military conquests (in geographical, rather than chronological, order), some other epithets and a description of the king's works at Kalhu with a focus on the construction of the NWP. A. K. Grayson counted 406 specimens of the inscription (RIMA 2, 274). We shall use here the same distinction between "text" and "inscription" used by Russell (1999, 7): "A 'text' is a verbal composition, while an 'inscription' is the physical result of replicating all or part of such a text in a durable medium." However, like Russell, we will make an exception for the Standard Inscription's text, for conventional reasons. art and cuneiform writing suggests that they were meant to be seen, though how explicitly they were to be understood depended on the context, and on the intended recipient of the interactions. With the aim of thoroughly analysing the question of the intended public, I will make use of a comparison with another Assyrian monument, the stele erected by King Esarhaddon (680-669 BC) at Sam'al/Zincirli (Turkey). This stele has already been analysed by B. N. Porter in a study on Assyrian propaganda, in which she noticed some examples of image-inscription interactions that will be useful for our purpose.
Finally, I will try to sum up some conclusions, with a reflection on the methodology that I think we should follow when treating image-inscription interplay and its possible meanings and/or functions in relation to other research questions, especially those focused on the functions of the reliefs in their architectural context.

identifying image-inscription interactions5
The main problem, when seeking to define an interaction between part of an inscription and the image that it crosses, is the possibility that the interaction occurred casually, and the observer's wish to identify it confers meaning where there is none. The very nature of multiple reading, which characterizes cuneiform writing and which (as we will see in the examples below) often makes an interplay between image and inscription possible through specific word games, can in other cases be misleading and lead the observer to an erroneous identification. Put more simply, the presence of a word "in the right place" is not a sufficient substantiation for the existence of the relative interaction with the image behind it.
We shall take as a negative example fig. 1, where it is possible to see the arm of a protective two-winged genie (apkallu) sculpted on slab S-a-2 (upper, right).6 At line 5 of the relief, the signs that form the word "arms" (i-da-at) are easily visible right on the genie's arm. These signs are taken from the phrase "When Aššur, my Lord, who called me by name (and) made my sovereignty supreme, placed his merciless weapon in my lordly arms".7 However, we may note in the first place the presence, on the same genie's arm, of the sign -na from the preposition ana, which precedes the word for "arms", and secondly we see the absence of any additional expedient, such as an empty space before or after the word, aimed at highlighting it. As a matter of fact, we lack any possible evidence that the word was deliberately put in that particular position.
Therefore, it is necessary to meet specific conditions in order to determine a visual interaction between image and inscription. Essentially, these conditions are two: the sign or group of signs have to be isolated or partly separated from the rest of the inscription by leaving empty space before and/or after it (but also above and/or below it); and the image element that interplays with the sign or group of signs has to encircle (and so isolate) it/them. Sometimes, the two conditions may occur together, acting as a further confirmation of the existence of the interaction.
Once the actual existence of the interaction is proven, the possible kinds of interplay between image and cuneiform sign/s will be worth considering. As may be seen in the table at the end of the article (Table 1, on p. 64-65), where all of the examples considered here have been summarized, images and inscriptions can interact in different ways: a. Possible de-contextualization of the sign or group of signs from the inscription to which they belong. a.1) The sign or even the group of signs can belong to a split word, i.e. only part of the word will be included in the interaction. a.2) Even when the word is entirely included in the interaction, the fact of being part of this interaction puts it in a new context, which can be (and usually is) different from that of the inscription's text. a.3) Because of the nature of cuneiform writing, one sign can be read in different ways depending on its syllabic or logographic values. A different reading of the sign is then possible but will necessarily depend, first, on its being isolated from the rest of the inscription and, second, on a sort of "indication" given by the image with which the sign interacts. The first example that we will see below of the "GIŠ on the tree" meets all of these conditions. b. Possible de-contextualization of the image. As part of the interaction, the image, or rather, the element of the image that plays with the inscription, can be extrapolated from its original context. For instance, in the interaction example noted by Porter Fig. 1: Relief S-a-2 from North-West Palace. In the frame the signs (a-)na i-da-at, "(in)to (my lordly) arms" are visible © Metropolitan Museum of Art, New York (elaboration by the author).
Unauthenticated Download Date | 1/30/17 12:13 PM in Esarhaddon's stele (see below), the king's sculpted feet "lie" upon a written text that describes them in the act of trampling on the rulers that did not surrender to Esarhaddon. c. The existence of a direct or indirect relation between image and sign/s. The interacting sign/s and image can have a direct relation, i.e. a relation in which the sign or word identifies the visual element on which it is written. The example described below from slab H-4, on which the name of king "Aššurnasirpal" is encircled by the hand of a sculpted king, would be, if the two were the same man, a perfect example of a direct relation. Elsewhere, when more complex reasoning is necessary to guess the relation between image and sign/s, we will have an indirect relation. The degree and quality of knowledge necessary to understand this indirect relation will then define the typology of the interaction and, to a certain extent, its intended audience.
The analysis of the above-mentioned characteristics allows us to study the possible meaning(s) and function(s) an interaction might have. In the following examples, we will see how in certain cases an interpretation of such meanings and functions was attempted. In other cases, such an interpretation was not possible, a fact that had the consequence of partly undermining previous theories.

A giš on a Tree
Slab B-23 is one of the most studied reliefs from the NWP. It is located on the eastern wall of throne room B, behind the throne dais, and it shows two kings, traditionally identified as two portraits of Aššurnasirpal II, facing each other with pointing fingers in the typical gesture of greeting.8 Behind both kings, in a symmetrical pattern, two human-headed apkallus are sculpted. Each carries a ritual bucket in its left hand, while with its right it points a pine-cone-like object towards the back of the king's neck. Both bucket and pine cone object are closely connected to the image of a stylised tree, which stands between the two kings as the axis of the entire scene: the so-called Assyrian Sacred Tree, whose image is constantly reproduced (about one hundred times) all over the palace's rooms.9 Above the tree floats a winged disc, and within it, a god greets the king on the right side of the panel, showing him the divine ring ( fig. 2). In this particular relief, the SI is written with evident care: the signs are clear, well-separated, and not indiscriminately carved on every part of the image they cross. The AST, in particular, is left almost completely untouched. This is not the only element of the relief to be free from signs (the same holds for the fringes of the characters' dresses, as well as the hands of the king and of the apkallu on the left side, and almost the entirety of the king's sceptre on the right side). This is, however, the only relief in the NWP in which the AST is left completely bare, with not a single element of it inscribed by cuneiform signs,10 except for one sign right in the middle of the tree trunk: a GIŠ, the Sumerian logogram for "tree" (fig. 2a).
As part of the inscription, the sign gets its syllabic value /is/, as the last sign of the participle mukabbis in the sentence NÍTA dan-nu mu-kab-bi-is GÚ a-a-bi-šú "strong 10 There are other ASTs that are partially left uncrossed by the inscription, but none of them is completely untouched by it. I have found four of them, see Meuszyński 1981, pl. 12, H-30, H-32;pl. 15, L-21 and pl. 16, N-2. Slab B-13 is an almost exact copy of B-23. Like B-23, it was set in a very important position, directly in front of the entrance D that led into the throne room from the external courtyard. The only noteworthy difference between the two slabs is in the iconography of the god in the floating winged disc, who faces left (= east) instead of right (= south; note that both gods' gestures point at the doorway that leads from throne room B to room F, see Brown 2010, 28)   In the black circle the sign GIš is visible. White circles: 1) sign GÚ, slightly deformed to fit into the space between branches; the same happens with 2) sign UGU, and 4) sign GAL. Sign Ú (3) was too long to fit into the spaces between the branches, and was therefore engraved in the first possible space before the end of the tree (photo and elaboration by the author).

Unauthenticated
Download Date | 1/30/17 12:13 PM male who treads upon the necks of his foes".11 The fact that the sign has been intentionally highlighted by the slab's artisans is apparent for three reasons: a. A clear equivalence of GIŠ with "tree". "Tree" is the first meaning of GIŠ as a logogram, and the same sign is used as a determinative for names of trees.12 b. In no other cases from the NWP (as far as I am aware) is a sign left as "isolated" on a relief as this one. As already mentioned, save for our GIŠ, not a single element of the tree is touched by the inscription. Looking closely at the slab, it is clear that even the space between the branches was filled only with signs, whose shape would not invade the area of the sculpted tree. There would have been, for example, not enough room to fit the Ú at line 12 until the last empty space before the end of the tree. Looking at certain signs, such as the GÚ of line 3 (the sign just after our GIŠ), the UGU (/ muḫ/, from the sentence a-na i-da-at EN-ti-a lu-ú it-muḫ13) at line 5 and the GAL (from the sentence É.GAL TÚG.MEŠ14) at line 15, one can even see how they were partially deformed so as not to touch (or at least, in the case of GAL, to touch only very lightly) the tree's branches ( fig. 2b). c. Among the variants of the Standard Inscription copies noted by Le Gac, one is the use of the sign SI instead of GIŠ.15 Needless to say (however forced this might be), no interaction between trees and signs are to be spotted in the corresponding reliefs.
In this example, then, the double reading of the sign reflects a kind of interaction in which the presence of a peculiar image (the tree) forces the reader to extrapolate a sign from its textual context (/is/) and to read it alone as a logogram (GIŠ), in direct connection with the image itself. Brown's interpretation of the meaning of this interaction starts from the Assyrian Sacred Tree. Different scholars16 have argued that the sacred tree-which very much resembles a palm tree (a feature that could put it in a close relation with kingship, see below)-served as an index that referred back to several characteristics of Assyrian kingship.17 On the one hand, however, recent studies have convincingly undermined the supposed equivalence between the AST and a palm tree,18 and on the other, it has been argued that the AST was probably not an actual tree, but a wooden and metal object that represented a tree.19 Consequently, Brown does not concentrate directly on the meaning of the sacred tree to explain the equivalence between sign and image. Instead, he considers the equivalence between the sign GIŠ and the name of the legendary king, prototype and index of the very institution of kingship: Gilgamesh of Uruk. As a matter of fact, as shown by Parpola, the simple addition of a pre-determinative for "god" (AN = d ) to the sign is enough to allow us to read the equivalence d GIŠ = Gilgamesh.20 Brown, then, suggests a double symbolism for this sign on the AST: the "prototypical warrior and builder king serving as a symbol of the larger institution of kingship".21 Before passing to the next example of interaction, I would add to Brown's conclusions only one consideration, because I believe his reasoning too complex, although I don't rule it out altogether. I would not reject a simpler equivalence of GIŠ = AST, even with the necessary prudence in doing so. Very briefly, the innumerable studies on the AST (one of the most popular subjects in the history of Assyriology) have now-thanks to Giovino's The Assyrian Sacred Tree: A History of Interpretations (2007)-reached a stage where we can posit a close interrelation between the sacred tree and the secrecy of kingship. The interpretations of what is or is not an AST, in any case, allow us to place it within a rather restricted symbolic area. It certainly does represent a tree, even if not a palm tree, nor a real tree. However, even if they are not directly linked to one another, the AST and the palm tree share a proximity to the king's sphere, since for example, as Richardson22 notes, the palm and the king (Sennacherib) can be defined with the same Sumerograms GIŠ.GIŠIMMAR, to say nothing of the many metaphorical relations between a tree (particularly the palm tree) and its seeds, and the king's offspring, which often occur in royal inscriptions. All together, these different theories do not put any relevant distance between "tree" (GIŠ) and kingship. Hence, I consider Brown's equivalence with the "divine tree" Gilgamesh plausible, but too indirect to be the primary one. I would say that the GIŠ on the tree is not the direct symbol of Gilgamesh, even though it shares with Gilgamesh (and one of the possible renderings of his name) the institution of kingship, in the same way that the AST is not a palm-tree, but shares with it the symbolism of kingship (as in Sennacherib's title). In a very similar way, I will admit that the AST was not an actual tree, but I do not see any reason, after so many convincing studies on this subject, to reject the idea that those who ritually saw, used, sculpted and finally recognized an AST on a palace relief would not have called it a "tree". In the same way, for thousands of years since the end of cuneiform culture, people in the western world have used the tree to illustrate genealogy-though at first the sole prerogative of kings and nobles, in today's 19 Giovino 2007, 177-196. 20 Parpola 1998, 324. 21 Brown 2010, 33. 22 Richardson 1999.

Unauthenticated
Download Date | 1/30/17 12:13 PM internet world, anybody can have his own digital "genealogy-tree", even though we know it has never been an actual tree.

A Name on a Hand (Aššurnasirpal)
Slab H-4 from the NWP shows a standing king performing rituals, and behind him, a two-winged human-headed apkallu. Brian Brown noted that on the king's left hand is inscribed the name of Aššurnasirpal (1aš-šur-PAP-A). It will be worth noting, also for the sake of what we will argue below, exactly how this is presented ( fig. 3). The Assyrian king (traditionally identified with Aššurnasirpal himself) is shown facing left (south), performing a libation ritual. With his right hand he raises a drinking bowl, while he keeps his left arm horizontally at waist-level, showing the back of the hand, which holds a bow from its upper end (the bow being held in rest position with the lower end on the ground, "opened" externally). The king's figure is sculpted very close to the left edge of the slab, and so it happens that the lower part of the king's left hand cuts across the sign of the very first part of the SI's first line, "Palace of Aššurnasirpal", consequently encircling Aššurnasirpal's name. No space is left before or after the hand; the inscription continues normally. It should also be noted that the very lowest part of the hand is crossed by the upper part of some of the signs from the lower line ( fig. 3a). Therefore, even if it is true that the king's name is clearly visible, highlighted by the presence of the crossing hand, I believe that it cannot be argued with absolute confidence that the scribe decided to highlight the king's name deliberately, by means of this particular image-inscription interaction.
Nevertheless, we will admit that the king's name clearly catches the eye, and it will be worth discussing its possible meaning(s). Brown's interpretation is directly linked to the main subject of his article, i.e. the interpretation of the reliefs and their relation to the functions of the palace's rooms in the sphere of the ancestors' cult. Brown rightly underscores the fact that this is the only relief in which Aššurnasirpal's name is interacting with an image in the NWP's reliefs. He also considers this detail while describing the difference, in particular, between NWP rooms G and H. As Brown notes, there is a strong similarity between all the kings depicted in room G, while in room H the kings are different from one another.23 Consequently, while the kings in room G can all be identified as Aššurnasirpal II himself (an identification that is also due to the functions of the cult that Brown posits for the room), those depicted in room H are different kings, probably ancestors. The name on the hand, then, identifies the only king that actually is Aššurnasirpal II or his eponymous Aššurnasirpal I, an identification that can become a substitution once Aššurnasirpal II dies.24 Notwithstanding the value of Brown's interpretation of the NWP's architectural functions, we will try to show in the next paragraph how attributing actual direct functions to image-inscription interactions can be misleading and risks undermining part of the general theory.

A Name on a Throne (Shalmaneser)
Slab 3 from room G is the central part of a three-slab relief showing a banquet theme. The king is depicted seated on a backless throne, holding a drinking bowl with his right hand, while his left hand rests on his legs. Left of the king (i.e. in front of him) stand a eunuch and, behind him, a winged human-headed apkallu. Similarly, to the right of the king (behind him) are two eunuchs and an apkallu with the same features. The scene is flanked by two other slabs bearing one large AST each. The inscription runs on a band that crosses the figures' bodies from under their waists to the upper half of their legs (the lower half in the case of the seated king). Each slab contains a complete copy of the SI ( fig. 4).
Looking closer at G-3's inscription, one can see how many parts of the image have been left uninscribed in the area around and including the backless throne ( fig. 4a). Considering the inscription as a horizontal band that crosses the relief from left to right, we see how it invades the left part of the king's figure, covering his bent legs up to the fringes of his clothing. Then, following the line of the fringes it stops, leaving intact the rest of the king's vest, his backside, the back drape of his clothing, the throne's cushion and its ornamental ram-heads. On the throne, the inscription runs along its side and legs. The space between the throne's legs is filled with an undulating motif that represents textile fringes coming down from the throne's side. This part is not inscribed, so that in this area the inscription is carved only on the throne's legs. To the right of the king and his throne, the inscription starts again, continuing uninterrupted to the end of the slab. There is one more distinctive trait: the last line, which consists of only eight signs, is carved after the right leg of the throne instead of starting from the left edge of the slab. Fig. 4a shows how the scribe roughly followed the curved line of the king's clothing, creating a smooth aesthetic effect, but was not as careful in the lower part of the inscription, where the lines are at throne-level. In comparing those spaces with the lines carved on the throne, we get the impression that he deliberately left the exact space that he needed to fit the signs he wanted. In this relief, we find two examples of image-inscription interactions, one on the throne's side and one on the throne's legs.
The side of the throne can be divided in two parts. An upper part corresponds to the seat, on which the cushion is laid. In its real appearance, this part was probably made of wood, while to its front and back metallic ram-heads were attached. All of the   together frame a lower part of the throne's side, formed by a strip of "wood" and the band of "textile" fringes hanging from it.
The "wooden" lower side of the throne is engraved with only nine signs ( fig. 4b), with a long empty space between them and the throne's right leg (the only part of the relief with some empty space on the right side of the inscription). If compared with the line above, this space looks large enough to have enclosed a further 6-7 signs. Furthermore, the space below the signs could have been filled with a further line, which starts on the throne's left leg but continues only on the right one. The thus isolated nine signs form the name of Shalmaneser and his epithet, "King of Assyria" (= [1] d šùl-ma-nu-SAG MAN KUR aš-šur25). As part of the inscription, the king's name refers to Shalmaneser I (1274-1245 BC), who is said to have been the founder of the city of Kalhu, rebuilt by Aššurnasirpal. Therefore, the name is highlighted in two ways: by encircling it in the frame of the throne's lower side, and by leaving some empty space to the right and below it. Here, there is no different reading of the signs when following the inscription or when looking at them alone. For its part, however, the role of the image is unclear. Is the name referring to the king seated on the throne? Does it refer to the throne itself, as a sort of property label? As far as I have been able to make out, this is the only other king's name highlighted on the NWP's reliefs, besides that of Aššurnasirpal on H-4. Now, as I mentioned above, the presence of Shalmaneser's name on this relief could undermine Brown's theory. If, in fact, the king on H-4 is given a name because he is the only one to be identified with Aššurnasirpal (I or II), shouldn't the king seated on G-3-the very room (and right on its "banquet" relief) that, as Brown convincingly explains, was decorated with only Aššurnasirpal's representations-be the only one not to be identified as Aššurnasirpal, but as Shalmaneser? And which Shalmaneser? The first, founder of the city of Kalhu, or the third, Aššurnasirpal's son and legitimate successor? My inclination tends to neither of these. This king is not Shalmaneser (I or III), nor can we have any certainty that the one on H-4 is or is not Aššurnasirpal. I find it similarly difficult to presume a functional relationship between bas-relief, room functions, and image-inscription interaction as the one suggested for H-4.
Rather, I would look for an ideological interpretation of these interactions that more broadly stressed the celebration of kingship, similar to what happens with the example of the "GIŠ on a tree". Besides deities and kings, the throne may belong to members of the royal family, members of the assembly, judges, and spirits of the dead.26 Furthermore, the throne often has a metaphoric meaning of firmly establishing kingship (kussī šuršudu).27 As suggested by C. Pappi, (and also keeping in mind the role of a king's ancestors in the NWP, as underlined by Brown in his analysis), this could be the ideological meaning behind the presence of the name Shalmaneser-the king who is celebrated as the prime founder of Nimrud in the inscription-on the throne of Aššurnasirpal II, his natural heir.28

Two Wooden Legs
The second interaction to note on slab G-3 is, admittedly, less evident than the one mentioned above, but it is rather interestingly repeated twice in the same way. On the right leg of the throne, fourth line from the bottom, and on left leg, second line from the bottom, we find the signs for the word "meskannu wood" ( giš mes-kan-ni) from the final description of the construction of the NWP: "I founded therein a palace of cedar, cypress, daprānu juniper, boxwood, meskannu wood, therebinth, and tamarisk as my royal residence and for my lordly leisure for eternity […] I hung doors of cedars, cypress, daprānu juniper (and) meskannu wood in its doorways" (fig. 4b).29 The meskannu/musukkannu (also mesukannu, meskannu, mismakannu, usukannu, Sum. m e s.má.kan.na, original meaning "mes tree of Makkan/Magan", possibly to be identified with the Dalbergia sissoo30) is a type of tree known to have been imported from Magan (presumably modern Oman). By the first millennium BC, the musukkannu tree also grew in the Near East, as we have attestations of Assyrian kings planting it (Aššurnasirpal II and Sennacherib), while Tiglath-pileser III (722-705) found a plantation of it in Babylonia.31 In the royal inscriptions, musukkannu wood is defined as iṣṣu dārû "lasting wood",32 and is known to have been used for the decoration of buildings and furniture of value, especially legs and bands.33 It is particularly noteworthy that the musukkannu is known as one of the valuable woods used for thrones.34 la SAA 10, 294: 10, "May they (the gods) keep the foundation of the throne of your kingship as firm as a mountain rock forever (išdi kussī šarrūtika kīma šipik šadî lišaršidu ana ūmē ṣâti)." 28 Pappi, personal communication. 29 RIMA 2, A.0.101.23, 18-21. 30 Gershevitch 1957, 317ff. 31 Postgate 1992. 32 See entry musukkannu on the CAD M II, esp. p. 238. See also for example the inscriptions from Tukulti-Ninurta II (890-884): RIMA 2, A.0.100.5, 70-73a "from Ilī-Ibni, governor of the land of Suḫu (I received as tribute): three talents of silver, 20 minas of gold, an ivory couch, three ivory chests, 18 tin bars, 40 furniture legs of meskannu wood, a bed of meskannu wood, six dishes of meskannu wood, a bronze bathtub, linen garments, garments with multi-coloured trim, purple wood, oxen, sheep, bread, (and) beer." 33 Salonen 1963, 221. 34 Pappi 2013. See also, from the Mari archives, the throne in musukkannum wood that Zimrī-Lîm Except for the sign -ni that "falls" out of the right leg, both words are precisely fitted in the "wooden" parts of the image and I believe that this position, with no inscribed space before and after the signs (except for the right side of the right leg), leaves room for the possibility that they were set there deliberately. If so, we would have here an interaction in which the word is kept (almost) in its entirety on the image, keeps its reading and is partly extrapolated from its original context, even though, when Aššurnasirpal claims to have built a palace of mesakanni wood, he might also mean its valuable furniture like the ritual throne. Moreover, if my interpretation of the aforementioned interaction between the throne and the king's name proves to be correct, a relation could be seen between the everlasting character of the meskannu tree and the ideological motif of kussī šuršudu, providing an interesting example of two interactions conceived around the same image, both aimed at conveying a similar message.

image-inscription interactions in the Frame of the NWP Decorative Project
The examples shown above need to be analysed in the broader framework of the construction of Kalhu and its palace. In particular, studies from the past forty years have permitted more detailed interpretations of the role of inscriptions (especially the SI) in the decorative project of the NWP. Besides their textual characteristics (narrative, linguistic, etc.), monumental inscriptions have a well-known symbolic and propagandistic meaning in their visual impact, fully aside from the ability of the viewer to read them. In its very essence, the inscription stands as a representation of the king's knowledge, and consequently, of his power and legitimacy.35 The visual effect provided by monumental inscriptions is particularly evident in the NWP, where each relief-slab had its copy of the SI, which is visually connected to the ones before and after it, giving the impression of a continuous text. This inscription formed a horizontal band about 1.5 meters high that either crossed the images or had a register of its own, dividing the reliefs in two separate scenes.36 A closer look at each slab clearly shows that, for the sake of this visual donates to Nagatmiš, king of Tigunani (north of Šubat-Enlil), to reward him for his victory against the king of Ašlakka (ARM 30,(424)(425)(426). 35 Russel 1999, 47: "the power of these inscriptions must reside not primarily in their content but rather in the very fact of their existence". 36 Most of the separated slabs had images of a narrative and historical type, such as battle scenes and royal hunts of lion and bull. These slabs were posted in the throne room of the palace (B) and in the rooms to the west of the inner court Y (WG, WH, WK / BB). Other slabs carried images that covered their entire surface and bore one of three different subjects: 1) the king performing different kinds of Unauthenticated Download Date | 1/30/17 12:13 PM effect, when a slab did not have enough space to contain the entire text, the inscription was truncated, often mid-sentence. This procedure has been shown to have been carried out in accordance with some criterion.
With the studies of Paley (1976), de Filippi (1977), Sobolewski (1982), Reade (1985) and Russell (1999), it has been possible to identify and analyse two versions of the SI. These versions, Type A and B, differ in few aspects, sometimes of possible chronological nature, as in the two ways in which the text defines the northernmost limits of Aššurnasirpal's conquests: "to the interior of the land Nirib" (adi māt Nirib ša bītāni) in Type A, and "to the land of Urartu" (adi māt Urarṭi) in Type B. Type B is thus dated to a later period, after the eighteenth year of Aššurnasirpal's reign, when he conducted his first campaign against Urartu.37 Following a reconstruction proposed by Reade (1985) and elaborated by Russell (1999), Type A, the earlier version, was carved on all or most of the unsculpted slabs in a period that extends from the king's ninth year (875 BC) to the campaign in Urartu. In this period, the sculpted slabs were still under construction. There are no known examples of Type A on sculpted reliefs, nor any cases of truncated inscriptions on unsculpted slabs. When the slab was too narrow to contain the entire inscription, another text with similar literary motifs, the so-called Palace Wall Foundation Text,38 was used. Somehow, after the Urartian campaign, and by the time the sculpted reliefs were ready to be inscribed, Type B had supplanted Type A and any other text. All sculpted slabs, in fact, are inscribed with Type B, no matter what size they are. When a slab is too narrow, the text is left truncated, often in mid-sentence, and no Wall Foundation Text, nor any other, is used to fill narrow spaces.39 This reconstruction, besides exhaustively explaining the distribution of the texts in the palace's inscriptions, suggests a period of time long enough for a group of royal scribes, entrusted with the composition of the relief inscriptions, to define in detail a work of great complexity and refinement, a work that could have included the opportunity to insert variations in the obsessive repetition of the same inscription, without corrupting the original text.
Sometimes, certain elements of the image were left untouched, usually the most elaborate ones (wings, fringes of clothing, bowstrings), which could be adversely affected by the addition of cuneiform signs, and on which, moreover, the signs would rituals (throne room B and rooms C, F, G, H, S); 2) protective deities with human or bird heads (present in all the sculptured areas of the palace); and 3) foreign people presenting tribute to the king (on the throne room façade D, ED, E). 37 For other differences between Types A and B, see Russell 1999, 30-41. 38 Russell 1999, 24-28. 39 Russell 1999 have been less readable.40 If we compare slab B-23 (the one with our GIŠ) and B-26,41 for instance, we see that the writing is placed indiscriminately on many parts of the latter (e.g. including wings and fringes), and is therefore far less legible than on the former ( fig. 5). 42 We will imagine these scribes planning the composition of the most important reliefs of the palace. It is also possible to imagine that they would have painted the SI's signs on the sculpted slab to see the possible outcome of the whole work before ordering the stone-cutters to start working.43 On the other hand, the rest of the orthostats were inscribed without particular care or distinctive traits, and in some cases they were even left truncated at mid-sentence if space on the slab was insufficient. Yet, even when truncating occurred, a possible criterion is visible, related to the sculptured subject. As Russell notes, on the narrow slab 11 in room G (1.12 m), where a king is carved, the SI is included in its entirety, with a visible overcrowding of signs. Close to G-11, slab G-9 is decorated with a simple courtier and is only 7 cm narrower, but has a truncated text.44 The creation of the palace reliefs appears to have been a highly complex compositional project, of which the image-inscription interactions we are dealing with here were but one feature.

Visibility and the Public
If the complex compositional project behind the interactions, and more generally behind the whole effort of the decorative plan of the NWP, is now clearer, a question remains: how visible were these interactions meant to be? Were they meant to be concealed in the complex figurative decoration, or were they meant to be seen clearly?
The concealed hypothesis seems to be supported by two substantial elements. The first is the overall difficulty in understanding some of these interactions, as even the most evident "GIŠ on the tree" needs a well-trained mind to understand it in its deepest meaning. The second feature is their inaccessibility, their location in the rooms of the NWP exposing them to a very limited group of people.45 Furthermore,   12 m), where the SI is complete (Russell 1999, 40) (taken from Budge 1914, pl. X).

Unauthenticated
Download Date | 1/30/17 12:13 PM sometimes their very positioning in the architectural arrangement made them even less visible. One should not forget that even the "GIŠ on the tree" was placed behind the throne dais, therefore behind the throne itself, leaving very few occasions for possible contemplation.
Nonetheless, one should take into account how Ancient Near East figurative art and cuneiform writing-the two intimately related forms of art involved in image-inscription interactions-not only show a predisposition for approaches of multiple reading. They are also conceived following arrangements that include the possibility of focusing the attention of the observer on one particular element of the (literary or figurative) work by graphically highlighting it.46 Ancient Near East figurative art appears, at least to our eyes, naturally suitable for double or multiple readings and multiple significance. It is evident at first glance that the ritual scenes sculpted on the orthostats of the NWP do not represent simple celebrations of single historically identifiable moments. On the contrary, they bear a number of different messages relating to many aspects of the cult and institution of kingship. Furthermore, such works of art are not conceived only to be contemplated. In fact, they are meant to involve their observer in a periodical (and in some ways perpetual) ritual.47 The presence of image elements characterized by multiple readings (the AST is a clear example) gives the bas-reliefs not only a quality of polysemy but also one of performativity, i.e. the possibility of doing things, or performing actions with and through them.48 46 As many scholars have pointed out, it is not possible to study figurative art without textual analysis, especially when they are two parts of the same work of art, i.e. the monument with its image(s) and inscription(s). In the introduction to their brief study on the codes of visual representations in the reliefs of the Assyrian king Sennacherib (705-681), Carlo Lippolis and Marco Benetti argue that only by considering the centrality of the bounds between visual art and cuneiform writing can we hope to coherently interpret Mesopotamian art. The two authors do not refer simply to the message of royal propaganda shared by the two forms of art, but especially to the way of conceiving visual representation as a written enunciation (Benetti/Lippolis 2011, 79). Their study develops a less recent idea proposed by Irene Winter, who, in her article on the royal rhetoric in Neo-Assyrian reliefs (Winter 1981), summarizes the main (physical, ideological, historical, etc.) relationships between Assyrian royal inscriptions and reliefs, starting from the case of Aššurnasirpal II's NWP. In particular, she indicates that writing and sculpted reliefs share, on the one hand, "a similar syntax of action and explicit articulation of action to consequence", and, on the other, "the same grammatical and compositional play on the ambiguity between the subject of the action / the king in the first and third persons" ( Winter 1981, 21). Furthermore, in a more recent article, she underlines the close relationship between the structural organizations of visual narrative in the reliefs and the very nature of the Akkadian language (Winter 1997, 362). 47 Nadali 2013. 48 The ritualistic property of figurative art is not limited to reliefs of mythological themes. For similar characteristics in narrative reliefs see Nadali 2001Nadali -2003. The subject of performativity of the Ancient Near East monuments is in many ways close to the subject of the agency they were to perform on the society in which they were conceived. See Marian Feldman's article on the Hammurabi codex-stela This active role of reliefs is also made possible through their graphic arrangement, which allows for different points of view for the same work of art, following a multi-centric principle. As we know, Assyrian reliefs were not conceived as static graphic representations of a given subject that reproduce reality following the principles of modern (western) perspective. On the contrary, in Mesopotamian reliefs, space is conceived and represented as non-perspective and multi-dimensional.49 This characteristic is particularly evident in the reliefs of narrative subjects, where different scenes are gathered on a single sculpted slab. Here, the different sizes of the elements do not depend on the fixed perspective of Renaissance conception (e.g. the closer objects appear larger). On the contrary, the principle that rules the size of the depicted elements is grounded in the spatial, psychological and temporal relationships between them.50 In the narrative orthostats from the NWP's throne room, for instance, we see how the size of the soldiers depends on their status, the role they play in the narrative, and their relation to a particular part of the scene. Looking at slab B-4 (lower register), we see that the assault on the fortified city can be divided into different sub-scenes placed around the city, depending on the size of the characters and their relation to time and space ( fig. 6). Close to the right edge of the slab, two officials, one archer and one shield bearer, have the same height as the city and represent the main narrative theme: the overwhelming attack conducted by the Assyrian army against the enemy. For the sake of dynamism, the enemies, shown at almost half the height of the two soldiers on the right, are proportionally too big if compared with the walls they are standing on. In the centre of the relief, two Assyrian soldiers are almost the same size (Feldman 2010), where she analyses the consequences of the introduction of perspective in spatial rendering of the sculpted scene, and the work on agency marked and agency ascribed in Mesopotamian monuments by Irene Winter (Winter 2007). Winter's article, in which she reviews Alfred Gell's Art and Agency. An Anthropological Theory (1998), also deals with the agency of language in monumental inscriptions, but focuses only on linguistic aspects and semantic relationships with the respective monuments' images. I believe that it would be worth adding to her conclusions a study on the active role of monumental inscriptions that included both their linguistic characteristics and their visual features. Such a study would start from their physical incorporation onto the monuments-a subject that Winter mentions in her work on the conception of royal image in Assyrian ideology ( Winter 1997, 74)-and would include the kind of image-inscription interactions we are dealing with here, in the light of social theory. 49 In their article, Benetti and Lippolis describe the non-perspective nature of Assyrian reliefs. Differently than in western art, where space is conceived following the principles of perspective, in the Assyrian reliefs there is no fixed point of view that defines the shape of objects, deforming them according to their position relative to the viewer. On the contrary, the objects are depicted "simultaneously", showing the relationships between them in space and time. As recipients of the work of art, we can appreciate the represented objects as very close-with minute, clearly visible details-and very far, as single parts of the whole (Benetti/Lippolis 2011, 100). For a critical analysis of western perception of non-western art (especially Ancient Near Eastern), according to the principles of perspective, see among others Bahrani 2003, 73-95 (but cf. Fales 2009), and Feldman 2010, 151-159. 50 Benetti/Lippolis 2011 as the enemies with whom they are directly interacting, as they try to pull a soldier down from the walls with a pair of chains ( fig. 6a). Still of another size are four Assyrian soldiers, one pair on top of a battering ram (shown as big as the city walls) and another pair intent on crushing the city walls with maces. The result of such multi-dimensional representation is that we are able to "zoom in" on the different parts of the battle scene, so as to better appreciate the narrated events.
Quoting the neurologist R. L. Gregory, Lippolis and Benetti explain this way of conceiving space in the Assyrian reliefs as a consequence of the very nature of visual perception, which is not passive and objective (as though seen through a modern camera). In fact, visual perception is the result of a decisional process based on limited sensory evidence, for which the brain dynamically creates an interpretation of what the eyes perceive.51 Moreover, the eye is naturally attracted to objects that are in different ways (size, position, etc.) more evident than others.
Cuneiform writing is also well known for its characteristic of multiple reading. The single cuneiform sign has two main readings, one (inherited from its Sumerian origins) as logogram, i.e. one sign = one (or more different) word(s), and one as syllabogram, i.e. one sign = one (or more different) phoneme(s). The example of the "GIŠ on the tree" shows this double nature well. In Mesopotamian culture, as has been explained by such scholars as Bottéro, the relation between the cuneiform sign and the signified realities is extremely deep, and it is bound to the notion that every element or phenomenon of reality (including writing) can be interpreted as a message 51 Benetti/Lippolis 2011, 83. Unauthenticated Download Date | 1/30/17 12:13 PM from the divine world.52 Under certain circumstances, the fact that two (or more) realities can be identified by the same sign (polysemy) or that two (or more) signs meaning different words can be read in the same way (homophony) can create substantial relationships. Furthermore, such relationships can sometimes-in specific texts of high literary importance-be actively modified for specific purposes. Similar word games, then, are not conceived on a merely linguistic basis.53 They are deeply rooted 52 Bottéro 1987, passim. 53 See, for example, a case from Esarhaddon's inscriptions (RINAP 4, 104 ii 2-8), where the god Marduk himself plays with cuneiform signs-in this particular case, fractions. After having punished Babylon and its inhabitants by designating a 70-year period of destruction (i.e. Sennacherib's devas- in the very idea of knowledge as conceived in Mesopotamia. As Glassner summarizes, "analogy was woven into the very nature of language. We are struck by the amazing deftness with which the Mesopotamians plied the resources of their languages to express the relationships between words and things. It is here that we touch on the core of their thought, for the analogical relationships were not only consecrated by words, they were founded by them".54 The hermeneutical importance of writing is also based on the fact that words and signs have, as figurative art, a quality of performativity. It should be noted here that, already in the Sumero-Akkadian royal inscriptions of Old Babylonian times, the pronounced and/or written word represented not only a descriptive means, but in fact the necessary condition to create a communicative relation with the gods. As Seminara points out, the king's (building or military) undertakings needed to be perceived by the gods through their senses, and writing had a central role in such ritual performance.55 In a recent article, I used this point of view as a basis to analyse the notably large number of references to sensorial perceptions (especially sight, hearing, and smell) used in the descriptions of landscape from Sargon II's Letter to Aššur. In the text, the Assyrian king (722-705 BC) writes directly to the main god of the Assyrian pantheon, giving an account of his victorious military campaign against Urartu in 714 BC.56 My opinion in that article is that such an abundance of references to sensorial perceptions in a text of high literary and cultic value has the performative function of facilitating the opening of a communicative channel between king and god during the reading of the letter.57 The performativity of writing, or more simply its feature of "doing" different things at one time due to the possibility of multiple reading, is also visible in examples in which a particular sign seems to be used specifically to attract the eye of the observer. We should note, for instance, Irene Winter's textual analysis of one of Gudea's royal inscriptions (22nd century BC). In Cylinder A, where the Sumerian king celebrates the building of a new temple for the god Ningirsu, Winter points out that after the end of the construction works, the temple is described with the words "it stood to be marvelled at". Here, a central role is taken by the Sumerian logogram U6, "marvellous", which is composed by two other signs, IGI, the logogram for "eye", and tation of the city in 689 BC) on the "Tablet of Destinies", the mighty god reconsiders his decision and reverses the order of the two wedges that form the number 70. In this way 70 years become 11, and Esarhaddon thereby finds the divine authorization to rebuild the ruined city without dishonouring his father. As Glassner has pointed out, "none of those operations was gratuitous; they were the very basis of Mesopotamian intellectual production." (Glassner 1995(Glassner , 1819. 54 Glassner 1995, 1818. See also Maul 1999 for the use of orthography and etymology as a hermeneutical method of Babylonian scholars, and more recently Selz 2013. 55 Seminara 2004, 537. 56 Thureau-Dangin 1912. 57 Morello 2013 É, the logogram for "house/temple". It would be difficult to deny that any sufficiently trained cuneiform reader would notice (and be attracted by) the use of such a composite logogram, which uses the very objects involved in the action of admiring: the temple and the public's eyes.58 In considering such characteristics of Assyrian figurative art and cuneiform writing, we will find it less difficult to understand the nature of the interactions between image and inscription. In this light, it becomes more plausible to suppose the existence of compositional projects, behind the reliefs from the NWP, that combine the possibility of putting certain words or signs in evidence, focusing (zooming) attention on a particular spot, and creating possible relations between those words or signs and the part of the image they cut across.
On the other hand, we should not overestimate this phenomenon and conclude that all written sources coming from the Mesopotamian world were bearers of this kind of polysemic double meaning. In fact, as Fales recently warned us, we will not forget that such "games" of multiple significance are confined to very few examples gathered from particular kinds of literary texts.59 At the same time, we will also agree with Fales on another aspect. When such examples of multiple readings occur, they seem to be the outcome of the presence, in the king's entourage, of scribes and scholars expert in the technical competences involved in the relations between the divine and human world-what Pongratz-Leisten has defined as Herrschaftswissen.60 Thanks to the correspondence of Esarhaddon and Aššurbanipal,61 we know these relations to have been particularly intense during the 7th century, but we can assume they were also at least partly extant in previous periods. Those belonging to this restricted circle, namely the king himself, his intellectual entourage and a few others, were the primary recipients, the intended public, of these complex literary and artistic operations.62 58 Winter 2007, 55-56. 59 Fales 2009: "the vast majority of Assyro-Babylonian written documents, i.e. those which for more than three millennia were concerned with administration, law, epistolary communication, international relations, and historiography, do not exhibit this extraordinary 'game' of multiple and multidimensional significance-and fortunately so, otherwise we would have never been able to piece together the skeleton of Mesopotamian history as we have painstakingly done for the last 160 years! In other words, these thousands of texts 'speak straight' to us almost 100% of the time-with terms, syntax and especially logic which we, some two thousand years after the last cuneiform tablet was written, have fully come to follow and understand, despite the innumerable linguistic difficulties and contextual obscurities involved." 60 Pongratz-Leisten 1999. 61 See esp. SAA 13. 62 The more refined and complex example of this kind of sapiential dialogue, conceived for the inner court only, is the one that we could also define here as the strongest example of interplay between figurative art and writing, in which the images are themselves words: the so-called "astroglyphs", small depictions of animals, anthropomorphic elements, objects of everyday life and astral symbols that, Hence, our conclusion on the image-inscription interactions coming from the NWP's reliefs will be that these complex visual-literary interplays, even if partly concealed by their position in the palace's spaces, were meant to be seen. Their ideal recipient would have belonged to one of the three categories indicated by Fales in his chart of "possible types of audiences for Assyrian bas-reliefs": the inner court (scribes, scholars, and other in-groups), gods, and the king himself. Consequently, the respective functions of the interactions were those of political-ideological internal propaganda (preaching to the converted), feedback for divine support (through the celebration of kingship and its bond with the divine world), and self-gratification of the king, in addition to his intent to preserve the memory of himself for future generations, especially princes.63 All this said, we shall take into account how very similar image-inscription interactions prove to have been used in a diametrically different context, for a different kind of recipient and with a different purpose: on a public stele, displayed outdoors and in a frontier territory. The stele was erected by Esarhaddon in the city of Sam'al/Zincirli Höyük (in the Anti-Taurus Mountains of modern southern Turkey), two centuries after the foundation of the NWP (fig. 7). The evidence provided by this monument, as analysed by B.N. Porter, shows a different use of image-inscription interactions, aimed at a much more direct, and in a way simpler, propagandistic message. The interactions briefly described below were meant for two possible classes of people: Assyrian officials based in frontier regions, and possibly, if we imagine the occurrence of a public reading of the stele, perhaps with the aid of a translator, a local non-Assyrian public.
The propagandistic functions of the Sam'al stele have been already extensively analysed by Porter in her article;64 here we will only describe the image-inscriptions as they appear on the monument.
The stele is 3.46 meters high and it was erected for public display in the gate leading to the citadel. It shows the king raising an emblem of royal power in one hand, while the other holds the sceptre and two ropes with which he tames two small captives, who are kneeling before him: they are the rebellious king Abdi-Milkutti of Sidon (defeated in 677 BC)65 and the crown prince of Egypt, captured during the Egyptian campaign described in the stele's inscription. through the filter of the aforementioned sapiential knowledge (especially astronomy), can be "read" or "translated" as names and well-attested epithets of the kings who commissioned them (see esp. Finkel/Reade 1996, Scurlock 1997and Roaf/Zgoll 2001. 63 Fales 2009, 281, Chart 1. 64 Porter 2003 This ruler has also been identified with Ba‛al of Tyre (Pettinato 1975). In the Sam'al stele, we see three different kinds of interactions: a. the sign -ia, suffix of the 1st person singular possessive pronoun "mine", repeated at the end of each of the first ten lines of the inscription, and highlighted by the space left to isolate it from the rest of the signs ( fig. 7a);66 b. the ligature aš-šur right in the middle of Abdi-Milkutti's beard, isolated by means of empty space left between the last four signs of line 13 ( fig. 7a); c. the positioning of the sentence "all the non-submissive, the kings who would not bow to him, like swamp reed he cut down and trampled at his feet",67 in the area of the stele under the king's feet ( fig. 7b).   In the first case (a), the lines are visibly and excessively spaced in the inscription area, so that the pronoun "mine" seems to stand apart from the other signs at the very end of each line. Here, the "interaction" is created by the absence of any image: the signs are isolated in the void (filled only by the ruling lines and by the signs themselves) that constitutes the background of the sculptured scene.

RINAP
In absence of a particular image, it is nevertheless possible to imagine an indirect relation between the sign ("mine") and the stele itself, as well as the place where the stele was displayed, the inhabitants of that city, and the two rulers at the king's knees.
In the case of the word "Aššur" on Abdi-Milkutti's beard (b), Porter points out that "the spacing of the signs in this line [i.e. 13] and the line before make it clear that this placement was deliberate […]. The ligature aš-šur was carefully isolated on the beard, separated from its pre-and post-determinatives [i.e. KUR, "land" and KI, "earth", both determinatives for toponyms] by space for a total of about three signs."68 Porter suggests interpreting this "Aššur" on the beard as an "ironic"69 property label, and she compares it with other objects taken as booty and similarly inscribed with a line of cuneiform labelling them as property of the king, certainly a plausible hypothesis. In this way, the ruler's portrait is enriched with a further detail that asserts his tamed condition.70 In contrast to the case of the GIŠ on the AST, here the word "Aššur", even if separated from its determinatives, keeps the same reading in the inscription and when extracted from its textual context and applied to the image of the captive king. The relation between image and signs is indirect because it needs "historical" knowledge to support it, suggested by the image.
In the third case (c), the interaction between image and inscription is not due to their intersection, but to their juxtaposition. Yet the carved feet "trample" on the sentence, as the king tramples on his foes. Here, the feet are extrapolated from their iconographic context to take part in a game with the words deliberately positioned below them.

Conclusions
The examples of interactions between images and inscriptions that we have analysed in these pages are all characterized by extraordinary techniques of elaboration-68 Porter 2003, n. 182. 69 Personally here, I do not see any irony nor mockery. Rather I take it as a plain and "objective" (with all its propagandistic consequences) declaration of property. 70 There may also be a similar interaction, although this is admittedly more speculative, on the Egyptian captive's head. Here too, the scribe left more than enough space between the last four signs of the line. In fact, there would have been enough space to fill the line with even more signs than in line 13. The only sign inscribed on the prince's head is a ŠÁR, part of the word AN.ŠÁR, again " (god) Aššur". through continuous acts of de-contextualization and re-contextualization-of the elements that coexist in the same monument. In the form of an inscription, writing acquires materiality, and it can be physically "moved", until it finds its place on the image sculpted behind it. Still, the complexity of this operation resides in the fact that inscription and image keep their normal reading, even when a new one is added, so that we end up having two (or more) texts for one inscription. This happens because of the possibility of multi-significance created by isolating particular words, but also by the different possible readings of the single sign. The image as well, or rather its part, is extrapolated from its original context to interact with the signs. We are able to see the creation of a new level of perception of the work of art, one that is not only visual or literary, but a combination of the two, a combination that produces a new message.
This complex technique of monument manipulation can then be applied to completely different contexts, which require different messages meant for different recipients. The stele of Esarhaddon at Sam'al has a clear propagandistic purpose, which is surely performed not only by the interactions between images and inscription. As Porter has pointed out, the analysis of this monument-and Til Barsip's possible comparison with monuments with the same iconographic theme-has revealed a complex project, partly based on information acquired from local tradition and political conventions, which is conceived especially for its audience. A certain kind of text, certain ways to depict the characters (their clothing and their proportions), and a certain place for the monument display, are all elements carefully manipulated for the desired purpose.
The case of the reliefs from the NWP is also the result of a complex project, but within a very different (inner) context. Here, we may observe a similarly careful choice in the reliefs' subjects, their arrangement within the architectural spaces of the palace, and their engraving with many different versions of the same text. However, the analysis of the interactions in the palace of Aššurnasirpal II has shown how, in contrast with the analysis of Esarhaddon's stele, we keep missing the functional relations between interactions, subjects of the reliefs and architectural spaces.
I would conclude with a very brief reflection, based on this lack of certainty. Especially in the case of the NWP, as we have mentioned in previous paragraphs, I believe we should reduce our attempts to interpret functional relations between interactions, reliefs and architectural spaces, and certainly we should avoid unambiguous solutions that could have the unfortunate outcome of limiting our research. In other words, as we understand that putting a GIŠ on an AST, even if it clearly creates a relation between the two, does not imply an equivalence between AST and "tree" (i.e. does not represent a sufficient datum for the debate on the subject of the nature of the AST), we will also avoid oversimplified identifications between names and kings. The name of Shalmaneser on the throne in G-3, or that of Aššurnasirpal on the king's hand on H-4, does not represent sufficient evidence to identify either the name of the depicted king nor the function of the room. As a matter of fact, what we have seen of Esarhaddon's stele shows that, even in that case, the "scope" of the interactions does not seem to be that of creating simple tags to indicate who is who on the monuments.
Rather, by leaving the task of decoding, case by case, the reasons behind the realization of these "games" of image and signs somewhat open, I would propose to take into account the apparently simplistic function of opportunity.
The reconstruction that we have seen of the phases of realization of the NWP's reliefs has clarified the fact that the decorated orthostats have been inscribed only with the Type B SI, and that this operation happened only after they had been sculpted in their entirety. The scribe or group of scribes entrusted with the realization of the complex project of composition had the opportunity to work with readysculpted reliefs and a rather brief fixed text. On one hand, the example of the GIŠ on the AST-on the very relief that would have stood behind the king's throne-appears quite similar to Esarhaddon's stele, in terms of how the general message of the relief given by the image "rules" the creation of its interaction with the inscription. On the other hand, in the reliefs of slabs H-4 and G-3, the depicted scenes seem to create a graphic basis on which it was possible to elaborate interactions, e.g. the king's throne and its legs, and the king's hand positioned at the right height to be where the SI would have begun. In these last two examples, the realization of interactions between images and inscriptions appears to have been the result of "simple" opportunity. As a way of conceiving and perceiving writing in which there is an interplay between elements originally not conceived to interact with one another, this produces variations on the main theme-the celebration of kingship-that are worth imagining, realizing and communicating, in a permanent dialogue performed in two directions: from the king to mankind, and from the king to the gods.
Visual interaction: The king's feet "lie" on the sentence. No space left.