The Maze Tableau from Tell el-Dab'a

The results presented here are the product of a collaborative effort over a period of three years in which M. Bietak has been a primary contributor.

      A. Iconography. The reconstruction of the Bull and Maze was dictated by a few key pieces which reveal connections between three distinct zones: (a) at the bottom go the half rosettes, (b) above them are the bulls and acrobats on a maze background, and (c) at the top there is a broad undulating red band. It is most likely that the painting was framed at its upper edge by inverted landscape. To the left of the maze, there was another scene which has a related but different subject: bull grappling.

The compositional structure of the Bull and Maze painting is Aegean, as can be shown by comparable scenes attested in glyptic art, most likely borrowed from mural painting.

In summary, there seems to be a connection between bull-leaping and the palace, a fact which can also be deduced from the fact that all the major entrances of the Knossos palace were decorated with bull imagery. If the maze represents a court in the vicinity of the palace, it was abbreviated in glyptic scenes as a grid ground line, and the purpose of the convention was to indicate palatial grounds.

      B. The computer reconstruction. The introduction of the computer was a genuine breakthrough in our work, both in restoring each single piece and in putting together the overall composition. The project presented here is a pioneer work in this sense.

A palette was first created, in order to keep the colours as close to the originals as possible. Photographs of the original pieces were scanned and existing colours and outlines were carefully detected and enhanced. Parts missing or badly worn are filled in by using existing ones as templates. The overall composition was pieced together with the aid of the computer. In the process, the grid pattern was also analysed and its logic was detected.

The computer technology ensures the utmost accuracy in the restoration, and facilitates our work greatly, for we could produce numerous variations of the overall composition with rela:tively little effort and time. Moreover, we can intervene and make alterations, if necessary, at any point in the future. It is a procedure which minimises subjectivity and arbitrary interventions, yet is at the same time flexible and highly efficient: in other words, a technology which provides excellent means for both the study of the fragmentary wall paintings of antiquity and their restoration.

N. Marinatos

A. ICONOGRAPHY

    I. The logic of the reconstruction

Although the Bull and Maze frieze with the bulls looks unique, to some even foreign to the Aegean, in fact it conforms to Aegean iconographical principles (Fig. 1). The structure of the painting exhibits a vertical duality: a space defined by the maze below and a red void above. The transition between the two realms is rendered by an undulating white stripe with a black border which most probably indicates the hilly horizon. In addition, there is a third element below the maze, which may be defined as a frieze consisting of half-rosettes. Two alternative rosette types, one more squat than the other, have been identified by C. Palyvou.

The reconstruction of the frieze is certain as far as its main compositional structure goes, and it was dictated by a few key pieces which reveal connections between three distinct zones. These are a half-rosette frieze, a maze background and a red void.

At the bottom go the half rosettes. Above them was a blue base on which the first row of bulls was placed. Pieces which included both the rosette and the blue base above it (F 34) gave us the data to restore the bottom of the frieze (Fig.2). Another important piece included the blue base, a bit of maze and the hind legs of a blue bull (F 24-101). This fragment told us that some bulls were galloping on that blue base. Another fragment was most important: an en face bull gallops to the left (F 4) against the maze. A yellow acrobat with a partially shaved head clings on to the bull's neck.

Some more fragments showed bulls partly against the maze and partly against red (F 107; F 122). The latter pieces gave us an idea of the upper part of the composition and proved that there were two superimposed rows of bulls. We could also tell that the maze ended in an undulating line, consisting of a black and a white stripe, above which there is a red void. We have several fragments with this undulating border, including one which includes the underbelly of a blue speckled bull (F 102). Finally, key piece F 48 should be mentioned because it depicts inverted landscape above the head of a blue speckled bull (Fig. 3). This bull was painted against red and could safely be placed on the upper part of the composition. It is most likely that the painting was framed at its upper edge by inverted landscape.

    II. Description and meaning

Four bulls are present on the Maze Tableau (Fig. 1. The bull on the left is not included). This is a minimal reconstruction, and we cannot exclude more animals. On the other hand, we do not have many pieces unaccounted for, which would warrant more bulls. As it is, two reddish brown and two blue speckled bulls have been reconstructed, two on each row. Three animals gallop to the left, but the fourth one, on the upper middle section of the composition, runs in the opposite direction, to the right. This is an anomaly that we cannot fully explain, but the direction is securely attested by the remaining reddish brown leg and blue hoof (F 100) against the maze.

There were at least three acrobats in the Bull and Maze scene. One has already been mentioned in connection with the en face bull F4 (Fig. 4). He is a young, yellow leaper with a partially shaved head and long locks. He clings to the bull's neck, and his fingers clutch the animal's skin. He wears a bracelet and a Minoan type seal around his wrist. His legs have been restored as being tossed upwards and the direction has been dictated by the upward position of the preserved torso. We have, in fact, a fragment of a leg with white leggings and blue shoes (F 12) which must have belonged to this acrobat, to judge by the style. Similar white leggings are attested on the Bull fresco from Tiryns (Rodenwaldt 1912, 162-165, pl. XVIII).

There is also a head painted against the maze (F 38), as well as a yellow arm with a blue ornament around the wrist. The arm is painted against the body of a reddish bull. The only way we could make sense out of the head was to render one acrobat as falling (Fig. 4). In this way, the head is against the maze. It should be noted that we have evidence of another fallen acrobat on a fragment which belongs to a subsidiary scene which will not be discussed further here. It is worth noting that the fallen acrobat does not have a partially shaved head but has a shaved sideburn.

The scene to the left has no maze as a background but an olive yellow ochre (Fig. 5). There seems to have been no half-rosette frieze below. A bull is broken down and held low by a man who rests his chin on the animal's head. In front of the bull, to his right a man confronts the bull, as though he is threatening him. This man is not on the same groundline as the bull; perhaps we are to imagine him on a second plane in the distance. Of his kilt only the upper section and a small fragment with a blue border are preserved. His footwear is conjectural.

The style of the bull teaser betrays a different group of artists from those who executed the acrobats of the maze scene. The scale is smaller and the head seems large in proportion to the torso and arms (Bietak 1994, 46). This confirms the hypothesis, also developed by E. Davis (here, vol. II), that several painters worked together on one and the same scene.

The scene to the left must have depicted events related to bull catching and grappling. Bull leaping was thus viewed as a continuation or general consequence of the bull grappling.

    III. Comparanda of the structure of the painting

The obvious comparison with the Taureador paintings from Knossos will not be discussed here. Suffice it to say that, although the Knossian paintings are later, they exhibit a great similarity with our frieze in the following respects: 1. There are several bulls there. In addition, there are several human figures, some leaping, some landing, and others hanging from the bulls' horns. Obviously the acrobats specialised in different tasks. Certain men distracted the bull and grappled with him, some leaped over him. 2. The individual Knossian bulls may not have been separated by individual panels, as M. Cameron's restoration suggests. Rather, there may have been two bulls in each panel as in the Tell el Dab'a frieze. For additional comparanda see L. Morgan's and M. Shaw's articles on the Tell el Dab'a and Mycenae paintings (Morgan 1995, 39-44; Shaw 1995 and 1996).

The compositional structure of the Bull and Maze painting is Aegean, as can be shown by comparable scenes attested in glyptic art. On the Tiryns ring (Sakellariou 1964, no. 179) a goddess and her demon servants are depicted in the lower part of the scene. She is to be imagined in an urban space, probably the palace itself, to judge by the presence of the throne on which she is seated and the half-rosette frieze which frames the scene. The urban/domestic space ends in an undulating line, beyond which there is vegetation as well as stars. Although the latter two elements are intermixed, we are probably to imagine the vegetation as being on a closer plane than the astral signs; perhaps the engraver implied that the plants are silhouetted against the horizon. The duality (urban space below, sky above) is thus present on this ring as on the Maze painting. The half-rosette frieze on both the seal and the Maze mural is surely not a coincidence, and shows that the same principles of composition were at work. I cannot imagine that a non-Minoan painter would have come up with this solution.

On a gold ring from Isopata, Crete, (Platon and Pini 1984, no. 51), we meet with a similar compositional scheme. On the lower section of the ring there are votaries (with possibly one goddess and one priestess). They are situated in a field of lilies. Above them is an undulating line, which, similarly to that in the Tell el Dab'a painting, indicates the horizon. Above the horizon is a descending goddess (as we can tell by the flying tresses). On the sky area are additional signs which pertain to the cult, the meaning of which eludes us. The two realms of this scene are likewise earth and sky or, to put it another way, proximity and distance. The same device, the undulating line suggesting a hilly horizon, is used. There are other glyptic scenes which employ a snake-like horizon to separate sky from earth. One such example is the golden ring from Mycenae (Sakellariou 1964, no. 17). We shall not discuss them further here. Suffice it to say that the compositional principle was most likely borrowed from mural painting. The Maze painting from Tell el Dab'a thus conforms to well established principles of Minoan compositions which were, at some point, adapted to glyptic scenes.

What the maze represents is a problem. In my view, it has nothing to do with the labyrinth or the myth of the Minotaur, and the only way we can decode it is by looking at contemporary art. The only Minoan parallel, the fragment of a maze painting from Knossos (Evans 1921, 356-357, fig. 256), does not enlighten us because it is without context. There are, however, examples from glyptic art which show bulls galloping or resting on a kind of grid groundline. This pattern is not completely identical with the maze; it is simplified as befits glyptic art. It is, however, close enough in conception to the maze to merit our attention. On a gold ring from Olympia (Pini 1993, no. 135) a galloping, charging bull is placed on a simplified grid pattern, a baseline with oblique vertical lines on its upper and lower sides. Similar scenes with galloping or resting bulls occur on a seal impression from Chania (Pini 1992, 145); a gold ring from Archanes (Evans 1930, 220 fig. 154; Younger 1988, 172 = Ashmolean no. AE 2237); and a seal from Mycenae (Sakellariou 1964, no. 91). What does the grid groundline signify? It is, of course, important to note the association with bulls. Obviously the seal or ring engraver wanted to indicate a particular kind of space on which bulls are either galloping or resting (broken down?). That the space corresponds to the maze pattern on our paintings stands to reason. The question is, however, what the space signifies.

What is certain is that the action takes place on official paved grounds, such as a court, and not in nature. If the grid groundline in glyptic art corresponds to the maze in painting, then the maze can only be a court with an official character; in short, it indicates palatial grounds (Shaw 1995, 91-120). One last argument to support the court hypothesis is that designs of Minoan floors, like one found by D. Levi at Chalara, actually did look somewhat like the Tell el Dab'a maze pattern (Levi 1967-68, 152-153, fig. 108). Courts and floors must have shared some features since they were both paved.

If the maze indicates a court, what does this mean and why was it necessary to show it on seals and rings? I would like to argue that the court is intimately associated with the palace.

In summary, there seems to be a connection between bull leaping and the palace, a fact which can also be deduced from the fact that all the major entrances of the Knossos palace were decorated with bull imagery (Marinatos 1993, 73-74; 1994, 89-93). If the maze represents a court in the vicinity of the palace, it was abbreviated in glyptic scenes as a grid groundline, and the purpose of the convention was to indicate palatial grounds.

    IV. Conclusions

The artistic conventions and compositional structure of the Bull and Maze painting are Minoan. The human figures, the details of the dress, leggings, partially shaved heads, jewellery etc. are elements that only an artist thoroughly versed in Minoan tradition would be able to reproduce. As far as the bulls are concerned, note that the patches are identical to those which feature on Minoan bulls and not on Egyptian ones. Finally, the rosettes are also typically Minoan. If artistic language is a code with well defined rules, the imitator will betray his origin by missing out the details which are self-evident to the artist intimate with the code. Such small points are the shaved sideburns, the characteristic patches of the bulls, the seal on the wrist of acrobat F 4.

The painting provides more information about bull leaping and proves beyond a shadow of a doubt that bull grappling and bull leaping were connected, and were associated with official space, i.e. palatial grounds. The latter are symbolised by the half-rosettes and the maze pattern.

 C. Palyvou

B. THE COMPUTER RECONSTRUCTION

The computer restoration of the Maze and Bulls composition involved two phases: (a) the study of the available material in situ, as well as the relevant comparanda; (b) the restoration of each fragment and the reconstruction of the overall composition with the aid of the computer.

The basic components of this composition (maze, rosettes, bulls, acrobats) are discussed separately.

    I. The Maze

The maze pattern depicted on a large number of wall plaster fragments consists of a linear repetitive pattern in black and red on a white background. The pattern is drawn along the lines of a grid which was executed on the wall plaster by horizontal and vertical string impressions (Fig. 6).

These fragments were carefully examined with the aim: (a) of detecting the underlying grid of the maze pattern and understanding the process of its design, and (b) of using this knowledge safely to reproduce a large surface of maze which could function as a base for the reconstruction of the overall composition.

It was quite obvious that the maze pattern was mechanically drawn along the string impressions, yet the procedure was very elusive. No two lines were parallel to each other and no two dimensions were the same. It was soon understood that the procedure was more likely to emerge by studying the string lines than the colour strokes. The work therefore focused on the string impressions: all pieces were carefully recorded by means of 1:1 copies on transparent plastic sheets, and photographs of all fragments were scanned into the computer.

It should be noticed that, despite the computer technology that was applied to this project, the concept of the maze pattern was understood while copying the pattern by hand; for it came as a surprise to see that it was not as easy to draw as it seemed. Gradually, however, a better way of doing it was employed until the 'easiest' way was found - and this revealed the whole logic of the pattern.

The main guidelines (string impressions) form a square grid 4.6 cm. x 4.6 cm. and are paralleled by secondary ones, both horizontal and vertical (Fig. 7). The painting of the maze is done in 5 steps, alternating between horizontal and vertical strokes. Each step is applied to the whole of the surface that is to be painted. The great advantage of this procedure is that it allows one to draw the maze in a blindfold manner, on no matter how large a surface, without losing control of the pattern.

The design most probably worked its way upwards and from left to right, as follows:

  Step 1 - horizontal black lines

1a: Starting from the first grid cross to the left, a horizontal black line is drawn extending to the right over the next grid cross and up to the nearest secondary vertical line. A new horizontal line, similar to the previous one, begins at the nearby grid cross, and this is continued to the end of the line.

1b: This is repeated on the secondary horizontal line above.

1c: Coming to the next basic horizontal line further up, the procedure is repeated, only this time the first line starts one box further to the right and the rest follows as before.

1d: This is repeated on the secondary horizontal line above.

1e: Moving upwards to the next pair of horizontal lines, the procedure is exactly the same but starting as in step 1a.

This continues upwards to the end of the composition and the result is alternating pairs of horizontal lines.

  Step 2 - vertical black lines

From the ends of all the horizontal lines short vertical lines are drawn, halfway down or halfway up accordingly. This needs no measurements for it is easy to evaluate the middle of the distance by eye. It is, however, one of the reasons for the minor deviations in dimensions attested throughout the design. Moreover, the oblique vertical lines observed in some cases are a further reason for inconsistency - though the obliqueness itself is not apparent because of the short length of the line.

  Step 3 - horizontal black lines

The ends of the vertical lines are joined with short horizontal ones in an alternating manner, and the black boxes are thus completed. This again is done by eye, and therefore no strict precision can be attested.

  Step 4 - vertical red lines

Within the black boxes the red double Ts are now drawn, starting with the vertical lines. Again, no calculations are necessary.  

  Step 5 - horizontal red lines

These are drawn last, thus completing the design. The red double Ts follow no string lines and are the sloppiest and most deviating of all lines involved.

    II. The rosettes

Fortunately, there are some large fragments of wall plaster depicting rosettes, some including parts of the maze as well. All fragments have string impressions following the same grid as the maze (they are the continuation of the maze grid lines).

The rosettes are based on half circles. After thorough examination of all pieces, it became clear that there were two types of rosettes. Their difference lies in the width of the middle zone and in the way the horizontal bands alternate in colour. The circles are not all well executed: some have been drawn freehand within the boundaries of the two horizontal blue bands and the vertical middle bands.

    III. Bulls and acrobats

There are numerous fragments depicting parts of at least 7 bulls. After a long and tedious examination it was concluded that all the bulls were of the same size and could have been painted with the aid of a model template. Such a template was drawn by the computer using all available fragments (Fig. 8). Several bulls from Aegean art were also scanned and brought to the same scale as the Tell el Dab'a ones; they all fit the model template quite well.

    IV. Introducing computer technology

The introduction of the computer was a genuine breakthrough in our work, both in restoring each single piece and in putting together the overall composition. Image processing technology offers immense potentialities, and it was only a matter of time before it was introduced to the field of wall paintings in archaeology. The project presented here is in this sense a pioneer work; the way this technology is used has resulted from a trial and error approach. What is important, we believe, is that the computer programme was adjusted and manipulated by the scholars themselves; there was direct feedback, in other words, between idea and action.

Our first concern was to creat a palette. In order to keep the colours as close to the originals as possible, we scanned tiny fragments of wall plaster with the basic colours and used them to create the palette (Fig. 9).

Next we scanned good quality photos of all the pieces using high resolution analysis. This enabled us to blow up the picture and examine minute details. Upon these scanned photographs existing colours and outlines of various forms were carefully detected and enhanced. The whole work is kept on transparent layers so that we can go back to the original at any time (Fig. 10).

Parts missing or badly worn are filled in by using existing ones as templates. The head of the en face bull, for example, was completed by copying the existing right part and flipping it over. The result was actually astonishing, for it matched all the traces that were discernible from that part of the head (Fig. 10).

The overall composition was pieced together with the aid of the computer. A large surface with the maze pattern was created first. The band with the rosettes was prepared in a separate file and brought to the overall composition. The five bulls and acrobats were also prepared in different files and added to the final composition by adjusting them to the maze (Fig. 1).

The benefits of the computer technology are immense: it ensures the utmost accuracy in the restoration - an accuracy that can be re-examined at any time, since all this work consists of different files and superimposed layers which can easily be dismantled and rearranged if necessary. This facilitated our work greatly, for we could move pieces around and produce as many variations of the overall composition as we could think of, with relatively little effort and time. And, of course, we can intervene and make alterations, if necessary, at any point in the future. It is a procedure which minimises subjectivity and arbitrary interventions, yet is flexible and highly efficient. It is a technology, in other words, which provides excellent means for both the study of the fragmentary wall paintings of antiquity and their restoration.

----------------------------------------------------

--------------------------------------------------------