| Illustration 1 |
Illustration 2 (and Cover) |  |
 |
Five years ago, a reconstruction of the full design had been proposed by Steven Cohen at the 6th ICOC in San Francisco, an abstract of his communication being available3but not the reconstruction itself. In the present article, a new attempt has been made to reconstruct the design by using common properties of the known fragments. These are the following: Recurrent elements; mirror imaging; overlapping. A second purpose is to comment on the question of how many carpets shared the design and gave rise to the existing fragments. Other problems -- date and location of manufacture, iconography of the design -- have not been investigated thoroughly and will be treated in a more cursory manner.
Pictures of the fragments (color prints, photographs, color slides) together with structural data were obtained from various sources (Table 1). Minor differences between reported sizes of the fragments were, when necessary, averaged. Knot counts were obtained for eight fragments (Nos. 1, 3, 6, 7, 8, 11, 12, 13) and were calculated in other instances by using good pictures and a magnifying glass. Where reported, the knots were Sehna knots open to the left. Structures of warps were available from six fragments (Nos. 3, 6, 7, 8, 11, 13). They were Z7S or Z8S cotton, natural white, two levels. Wefts, if mentioned, were cotton, more detailed information missing except for two instances (Nos. 3, 13; Cotton, 4Z, 3 shoots). To facilitate reconstruction, line drawings were executed of the 15 fragments. Recurrent major fractions of the design were named A, B, C, D and were supplied with indices (R, L = right, left), depending on their locations on the mirror images. For convenience of presentation, fractions A, B, C, D were considered as "normal" when the (downward) orientation of the Sehna knots was in agreement with the orientation of the design as depicted in Illustration 3; in other words, the Sehna knots had to point towards the bottom of Illustration 3. Fragments, where the above orientation of the design (shown in Illustration 3) would result in upward orientation of the Sehna knots, were regarded as "upside down". According to this classification, fragments 1 through 5 had "normal" orientations, and fragments 9 through 15 had "upside down" orientations, whereas the orientations of fragments 6 through 8 remained unknown. No fragment gave the faintest hint of a mixed orientation (mixed = "normal" and "upside down" occurring within the very same fragment).
 | Illustration 3 |
1. Fractions A and B: At first sight, they seem to be slightly asymmetric halves of one common fraction. The designers, however, have treated them as separate units, exchanging their order of occurrence in quite the same way as with the other two fractions (C, D). Minor differences occur between different fragments, concerning the order of origination of animals out of the lion's mouth in Fraction A (see Illustration 4), resulting in three subgroups of fragments: Nos. 1, 2; Nos. 5, 7, 8; and Nos. 9, 12. A major deviation is found in fragment 7 (Illustration 4), where the quadruplet of the heads of two lions and of two antelopes appears below the parrots in a position which should be occupied by the elephant.
 |  |
2. Fractions C and D: Both are juxtaposed horizontally, but -- as in A and B -- the order changes depending upon the level of occurrence. This is shown by comparing fragment 3 and the lower part of fragment 1 with the upper parts of fragments 9, 10 and 12 (Illustration 4). The latter fragments show the sequence C,D,D,C, while the former suggest a sequence D,C,C,D.
Horizontal dimensions of C vary from 0.85 to 1.00 meters, vertical dimensions are 1.10 meters or less. For D the respective dimensions are 0.75 to 0.92 meters horizontally and 0.70 to 0.87 meters vertically.
3. Flowers, Shrubs, Vases: Interspaces between horizontal levels (A,B or C,D) are filled with flowers and shrubs. They never make contact with the animals. The interspaces vary considerably in height (interspace between A,B and C,D is 0.50 meters in fragment 1, and is zero in fragment 9). These differences limit the number of fragments which could possibly fit each other horizontally. Vases always appear upside down as compared to the orientation of design as shown on Illustration 3.
4. Reconstruction of Design and Number of Carpets: Looking at Illustration 4 and taking for a guide line the preceding paragraphs, the reconstructed design should have, from bottom to top, the following sequence (see Illustration 5):
Lowest level: DL, CL,CR,DR
Second level: AL, BL,BR,AR
Third level: CL, DL,DR,CR
Uppermost level: BL,AL,AR,BR
The uppermost level needs further explanation: AL and AR have elephants, BL and BR have not. Therefore, the combined fractions AL,AR are a combination with elephants on either side. What has been done with the four fractions is that, on their next respective level of occurrence, A,B have been converted to B,A and C,D to D,C. Illustration 5 shows the reconstruction. Its dimensions are 3.40 to 3.80 meters horizontally and, approximately, 5.10 meters vertically, including the separated top of the reconstruction (1 meter). This reconstruction was performed by using fragments containing more than one of the above fractions of design or subfractions, respectively. Illustration 4 shows the relevant pieces.
Illustration 6 shows fragments with opposite design orientations and their positions on the reconstruction. If Illustration 5 represents the design of a complete carpet's field, overlapping fragments with identical orientations -- for example Nos. 9, 11, 14, 15 (see Illustration 6d) -- must derive from different carpets. Thus, the total number of carpets would be six a minimum (as an be calculated from Illustration 6 and making allowance for the two possible design orientations of fragments 6 through 8). Following the style of execution and dimensions of the figures, some fragments seem to belong to each other: Nos. 2 and 5; Nos. 1 and 4; Nos. 9, 10, 13; Nos. 14 and 15 and, probably, Nos. 7 and 8.
 |  |
The reconstructed field design perfectly matches 14 out of 15 fragments, while one fragment (No. 7) exhibits a deviation from the common scheme, which was, however, not regarded as significant. The design of which Illustrations 5 and 7 each show sections is a drop design on the long axis. Its right and left halves are mirror images. The design is composed of four different fractions (A,B,C,D) depicting strange groupings of animals. Each fraction has, on the one side, the half face of a faun which matches up to a full face when two neighboring fractions are conjoined, but not, however, in the case of mere juxtaposition. Every two fractions belong together. On a given horizontal level, they appear as a pair together with its horizontal mirror image. On the next horizontal level above or below, the other two fractions appear in quite the same way. On the third level, the first pair reappears, but in a reversed order of its constituents, and the same happens to the second pair on the fourth level. Following this (on the fifth level), the cycle is resumed. Thus, a sequence beginning with A and B at the baseline would have the order: ABBA, CDDC, BAAB, DCCD, ABBA, etc.
Regarding the underlying carpet or carpets, the width of the field seems not to have extended over more than one width of the design (3.40 to 3.80 meters). The length of the field had to be a section of the drop design, as will be discussed below. Concerning the carpet border, no data are available.
The vertical orientation of the design is ambiguous; in other words, one cannot determine what is the top and what the bottom. The choice is not made easier by relating the design to the orientation of the Sehna knots, since fragments exist with more or less identical sections of the design but with opposite orientations of the Sehna knots as related to the design (Illustration 6). Should the two different kinds of fragment have derived from one single huge carpet, its upper and lower halves would have shown the two halves of a vertical mirror design, its total field length having been more than 15 meters. No fragment was found to show a transition from the one design orientation into its opposite. Furthermore, vertical dimensions of the above fractions (A, B, C, D) differed by more than 20 percent between the two kinds of fragments.
These findings were used to conclude that at least two carpets gave rise to the fragments, with opposite orientations of the design and with approximate minimum lengths between 6.50 and 7.00 meters. The field lengths probably covered a little less than two complete cycles of the drop design. In case of even more than two carpets, minor sections might have been chosen for a whole carpet field. Unfortunately, structural data are incomplete. Knot densities, available from 14 fragments, varied from 2450 to 3500 per square decimeter which, in worn fragments, does not contradict a common origin, even less so, since major variations occurred in one and the same fragment. Structures of the warps were reported in six instances, being cotton Z7-8S consistently. Finally, structural data neither supported nor contradicted any of the above considerations (which they would have done, if major differences would have been reported).
The first of the red ground fragments to be published was the Jeuniette fragment, actually two adjoining fragments, now in the Louvre (see Nos. 14 and 15). This was in 1907, the attribution was to India14, which is in agreement with present opinions. Following Martin12 who published the second fragment in 1908 (No. 9, the Roden fragment, now Detroit), the two fragments had appeared on the Paris market a couple of years earlier. No information is available concerning their previous fates.
Indian manufacture is suggested, more or less, by certain elements of the design such as parrots, elephants, a pink umbel on a wine-red ground (No. 13). Z7-8S cotton warps are found in many Indian carpets, some even from Jaipur. Their datings have become a matter of controversy in view of the possibility that multi-ply warps are not handmade6. Tentative datings for the red ground fragments, however, go back as far as the end of the 16th century17, although even an early 19th century dating could be considered.
A complete carpet with the design under discussion was, and probably still is, in the Bernheimer collection1. It is, however, executed on a smaller scale and in a very different style, more similar to 19th century vaq-vaq (talking tree) renderings. Nevertheless, it gives a complete picture of the above cited Roden fragment, which is the biggest red ground fragment showing this section of the design; but, additionally, it includes details outside the lower end of the Roden fragment. Thus the designers of the carpet from the Bernheimer collection seem to have had better knowledge of the design than was available at the time, when the red ground fragments appeared on the market. Judging from the style, they used another cartoon -- maybe in another country and at another time?
Where does the design come from, and what is its meaning? The vaq-vaq has been stressed in this context and has correctly been rejected. The design of the red ground fragments is not a tree giving rise to animals as a kind of fruit or leaves. It is not even tree-like. In fact, the basic scene shown on the fragments is not comprehensible. Are the small animals being devoured by the bigger ones, or are they born from the latters' mouths (a fecundity concept, as believed by Ellis7)? While there are book illustrations with the vaq-vaq which could have been copied by carpet designers, no such illustrations have so far been recognized with the design of the red ground fragments. Concerning single beasts or the fauns, they seem to have been internationally employed in the arts of the 17th and the 18th centuries and even later. Horned demons with serrated leaflike eyes, as in the fauns of the red ground fragments, are found in 15th and 16th century Persian and in early Mughal book illustrations, as well.
Summarizing the data, the design depicted by the red ground animal grotesques fragments is a mirror design on the horizontal plane (right-left) and a drop design on the long axis with a cycle of five consecutive horizontal levels. Fragments with identical designs occur as two groups, each upside down compared to the other. They probably derive from two or more different carpets. Their origin is commonly agreed to be India. The time of manufacture could be, from the structural point of view, as late as the early 19th century, whereas, stylistically, a 17th or 18th century date seems more probable. The roots of the design could have originated in 15th to 16th century Persia without the design itself occurring in early book illustrations. But it seems to have been well-known and to have spread across the times and countries into entirely different styles. So far, its meaning has not been understood.
1. Bernheimer, O., Alte Teppiche des 16. bis 18. Jahrhunderts, der Firma L. Bernheimer, Fig. 78, O. Bernheimer Ed., Munich, 1959.
2. Biriukova, N., Decorative Arts in the Hermitage, Plate 50, Aurora Art Publishers, Leningrad, 1986.
3. Christie's, UK, Rugs and Carpets, Auction Catalogue 10/94, Lot 569.
4. Cohen, S., "The Red Ground Grotesque Fragments," abstract ICOC, San Francisco, 1990, published D.R.Dodds & Assoc., Inc., Philadelphia, PA.
5. Eiland, M.L., Jr., Chinese and Exotic Rugs, Plate 43, A. Zwemmer LTD, London, 1979.
6. Eiland, M.L., Jr., "Some Unresolved Technical Issues," Oriental Rug Review III/11, 1983, pp 312-314.
7. Ellis, C.G., Commentary on Illustration 192 in Martin, F.R., A History of Oriental Carpets before 1800, Wien, 1908, Reprinted Oriental Rug Review, V/12, March 1986, p. 17/553.
8. Folsach, K.v, Islamic Art, The David Collection, Copenhagen, 1990, Plate 418, Ed. Davids Samling, Copenhagen, 1990.
9. Gans-Ruedin, E., Der indische Teppich, p. 70, Bussesche Verlagshandlung, Herford, 1984.
10. Hali 27, 1985, p. 59.
11. Hali 67, 1993, p. 100.
12. Martin, F.R., A History of Oriental Carpets before 1800, Wien, 1908, reprinted Oriental Rug Review, V/12, March 1986, p. 14/550.
13. McMullan, J.V., Islamic Carpets, p. 44, fig. 8, Near Eastern Research Center, Inc., New York, 1965.
14. Migeon, G., Manuel de l'art musulman II: Les arts plastiques et industriels, Illustration 359, A. Piccard et Fils, Paris, 1907.
15. Musée du Louvre: Arabesques et jardins de paradis. Collections Françaises d'art islamique, p. 142, pl. 112, Ed. M., Bernus-Taylor, Paris, 1989.
16. Sarre, F. and Martin, F.R., Ed., Die Ausstellung von Meisterwerken Muhammedanischer Kunst in Munchen, 1910, Nr. 180, Pl. 84, reprinted Alexandria Press, London,1985.
17. Welch, St. C., India, Art and Culture 1300-1900, pp. 159-160, fig. 95, The Metropolitan Museum, New York, Holt, Rinehart and Winston, New York, 1985.
