A White Central Asian Asmalyk
of a Rare Type

This article is dedicated to all those Turkoman collectors and specialists who appreciate the importance of those seemingly small and insignificant aspects of Turkoman weaving which are significant to the proper study of them.

The Bachelor's Hall Asmalyk Size: Width, 4'6" (137cm); Height, Sides 1'10" (56cm), Top 2'9.5" (85cm) Colors: White, red, salmon, deep blue, brown Warp: Wool, white, Z2,S Weft: Wool, white and brown, Z2,S, alternating colors in each shed with white first; the brown wefts have a larger diameter than the white wefts and are more visible Edges: Overcast in red wool, Z2,S, loosely plied like the pile threads, around three warps; the wefts interact and return only around the inner two edge warps Knot: Asymmetric, open left, no depression; no special edge knots Count: Horizontal 10, vertical 12, 120 per square inch; measurements taken in different places; slightly lower knot count in the outer 2" on each side Pile: Woven bottom first, clipped short; pile threads of white, brown and blue are Z2,S and red and salmon are mostly Z1 Kilim Finishes: Wool, white wefts, bottom 1-3/4" (4.5cm), Top 1" (2.7cm), half turned under and sewn down with brown wool. Condition: Very good with only minor losses to the overcast edges, some breaks in the brown wool thread used to stitch down the top and bottom kelim finishes and minor losses to the kilim in the lower left corner and point at the top.

This piece was auctioned by Rippon-Boswell in Wiesbaden in their sale of September 28, 1993, lot 121. It was described as a "Saryk," asmalyk style, old, Central Asia, Afghanistan, beginning of 20th century, and was estimated at 3,200 DM. It sold for a hammer price of 3,000 DM (c. $2,000). The Saryk attribution in quotations indicated Rippon Boswell's feeling that Saryk was questionable. The piece was initially accepted based on a photo because it looked comparable to a piece they had auctioned several years earlier. (See below.) When it arrived, they felt it was considerably younger and different from that piece, hence its date and estimate.

The piece was consigned by an American who had received it from his in-laws. It had come to them years before from grandparents, who had come to the U.S. at the turn of the century from Russia. It was discovered in the in-laws' attic, wrapped in paper, among items from the grandparents which had never been used after they were inherited. It is possible the piece had been wrapped and stored for at least 50 years. Measurement and technical analysis of the piece show the following features:

Because the piece was attributed to the beginning of the 20th century, a sample of the red wool was submitted to H. Schweppe for analysis. His results showed it to be the synthetic azo dye Fast Red B. Schweppe reports it was discovered by H. Baum in 1878 and named Acid Red 17. Other commercial names for the same dye are Acilan Bordeaux B, Lithosol Claret B, Acetacid Red 3B, and Kiton Bordeaux G.

The presence of a synthetic red provides a date certain prior to which the piece could not have been woven. It was woven sometime after 1880 and the early 20th century attribution of Rippon-Boswell could be correct, although the family history data implies a slightly earlier date.

This information was interesting because the other two pieces are attributed to the mid-19th and 18th centuries. It appeared that a comparison of the three pieces might prove instructive on several issues.

A full technical analysis of the piece was not published in either of the publications. A close inspection of the piece as published shows the following features:

Size: Width, 4'5" (135cm); Height, Sides not given, Top 2.9" (84cm)
Colors: White, red, salmon, dark blue, brown
Warp: Wool, white
Weft: Unknown
Edges: Overcast in red wool
Knot: Symmetric
Count: Unknown
Pile: Woven bottom first, clipped short
Kilim Finishes: Wool, white wefts, half turned under and sewn down with white wool(?) thread; top edge worked with red and blue wool.
Condition: Very good with only minor losses to the overcast edge, one small worn spot on right edge and minor breaks on top edge.

Oriental Rug Review and Hali both commented on this piece in their auction reports:

ORR: "Agreement ended at rare and important; opinion as to its trial attribution was another thing. 'We heard as many attributions as there are Turkoman tribes,' said one Sotheby's person. Jay Jones....told us he was able to touch the piece in passing and that it had a handle that suggested Beshir to him. Others told us they thought it was Saryk. When Dr. Jon Thompson....entered the lists, speculation on an Imreli attribution buzzed around the room." November, 1981.

Hali: "A high point...was a splendid symmetrically knotted white ground asmalyk... Apart from its beautiful and hitherto unknown design, its appeal was also based on its attribution by Turkoman specialists as the first Saryk asmalyk to have been discovered." Vol. 4, No. 3, 1982.

I recall inspecting the piece during the preview and finding its handle to be more akin to Ersari than Saryk rugs, although the Turkish knot was puzzling. Tom Baker recalls thinking that the red was too evenly saturated to be vegetal.

The Rippon-Boswell Asmalyk Size: Width, 4'3.5" (131cm); Height, Sides 2' (61cm), Top 2'11" (89cm) Colors: White, red, salmon, dark blue, brown and possibly a mid blue Warp: Wool, white Weft: Unknown Edges: Overcast in red wool Knot: Symmetric Count: Unknown Pile: Appears to be woven top first Kilim Finishes: Wool, white wefts, half turned under and sewn down with white wool(?) thread, top edge worked with red and blue wool. Condition: Catalog says full pile and good condition with no restoration to worn area in peak; minor stains.

A full technical analysis of the piece was not published, but from the illustration the following features are discernible:

Oriental Rug Review published the piece in color in October/November, 1990, p. 52, but with no additional factual information. Hali wrote, "...the Saryk attribution is still not universally endorsed -- there are some Yomut features according to our Turkoman consultant." (Vol. 13, No. 1)

Size. The sizes are quite comparable and the minor variation is standard for pieces of like types:
BH 4'6" (137cm) x 2'9½" (85cm) x 1'10" (56cm)
S 4'5" (135cm) x 2'9" (84cm) x 1'10" (57cm) (estimate based on image)
RB 4'3½" (131cm) x 2'11" (89cm) x 2' (61cm)
The BH and S pieces are closer in all dimensions than the RB piece, which is narrower and has a higher rise from the side measurement to the peak. The difference in the rise of the peak between the RB piece and the BH and S pieces is obvious even in the printed photographs.

Colors. All of the pieces have the same number of colors. Although printed color images are typically unreliable, it appears that the RB piece may have a lighter blue than the other two. The intensity of the red and salmon are stronger in the BH piece than the S and RB pieces. The RB piece has the lightest coloration of the three. In all three pieces the white has "yellowed" with age. Although no history on the use of the S and RB pieces is known, it could be assumed that they may have had more light exposure than the BH piece thus producing the more "faded" red and salmon in those pieces. A close look at the browns in the RB pieces shows some of the browns to be much lighter than the other two pieces, a reaction of light on natural brown wool.

Warp. Since Turkoman weaving is characterized by Z2,S spun warps, it is reasonable to assume all of these are so spun.
Weft. The use of alternating white and brown wefts in a Turkoman weaving is unknown to me in any other piece. This is one of the critical missing comparisons with the other pieces. It is known that regular alternations occur in pieces with cotton and wool wefts and in Eagle Group pieces with silk in the wefting. The reason for the use of this alternation is more difficult to know. On a white ground pile weaving, the pattern may have been easier to follow with a brown weft against which to tie knots. Being more loosely plied, brown wefts may have been used to produce greater pliability. There may be other reasons.
Edges. The use of a red wool overcast is common to all three pieces. Red overcasting is uncommon in Turkoman weaving, and its use shows a strong relationship between the three weavings. It would be useful to know if the same interaction of weft with three warp threads of the edge is common to all three as well. Knot. The BH asmalyk has an asymmetric knot open to the left. The other two are woven with symmetric knots. This is a major difference. The Saryk attribution for the S and RB pieces is largely based on the use of this knot. However, Saryk attributed pieces are known with asymmetric knots entirely or mixed with symmetric knots. This in and of itself would not seem to deny a common Saryk origin, if they are Saryk.
Count. The knot count of the BH piece is low for a Saryk weaving. The comment reported in Oriental Rug Review that the S piece "handled like a Beshir" indicates that it too has a low knot count and floppy, pliable handle. A comparison of the 120 knots per square inch of this piece with 10 pieces recorded in Turkmen show it to be lower than any of those. Only one of these 10 is close at 123 knots per square inch in a main carpet. In fact, in most Saryk weavings the vertical to horizontal ratio is 1:1.5-2.0 instead of the nearly equal ratio of 10 to 12 of the BH piece. Similar relationships are recorded in the analyses on Saryk weavings in Turkmen, Azadi, Wie Blumen in der Wuste, and Hoffmeister, Turkoman Carpets. This seems to mitigate a Saryk attribution.
Pile. The short pile and loose weave are unusual. With any amount of wear, the wefting would have become visible, particularly with a white ground and brown wefts. The use of differently spun and plied threads seems unusual, but there is not enough data recorded on this technical aspect of Turkoman rugs to make a judgment. The use of three-ply pile yarns in Eagle Group pieces shows that two-ply pile threads was not overriding. The significance of this feature is difficult to assess.
Kilim Finishes. The bottom and top kilim finishes of these pieces is common to most Turkoman asmalyks. The use of either white or brown threads to sew them down has been noted in most Turkoman asmalyks of all groups.
Condition. All of these pieces are in very good condition. The fact that the BH piece was in storage for so many years undoubtedly accounts for its condition. The other two pieces could have been similarly preserved. However, most rugs of great age have seen use and typically show more use than that present in any of these three. The condition could lead one to conclude that it is due to their having been woven at the end of the 19th century rather than having been preserved because of their special importance in Turkoman society. Most Tekke bird and animal/tree asmalyks show wear from use in the West, if not in Central Asia, and the oldest ones are more worn than the younger.

The S and RB pieces have what appear to be twined red and blue threads in the top edges. This is a fine finishing detail which is both decorative and functional. Twining may be commonly seen as the first color insertions in the skirts of main carpets of the Yomud, Ersari, and Tekke, where they are complete, so this is not unknown to Turkoman weavers. Two-colored twining is found in most bags and trappings of the Ersari. In Saryk bags and trappings, they used a single color, usually red, which was wrapped rather than twined.

The rosette kelles are carried on red stems and the stems of the bottom pair of branches are mostly dark blue. The center of the rosette contains a dark blue diamond. The star kelles have paired branches with red stems. The center of the particolored stars has a white center.

The exact drawing of the kelle in these asmalyks has been found in no other published Turkoman weaving1. There are similar ones in Saryk, Salor, Tekke and Ersari bags, ensis, and rugs but nothing identical. One must conclude, until other pieces become known, that the drawing of these two kelle forms is unique to these pieces.

The pattern of the two minor borders is called tumar or dogashik. A tumar is a triangular-shaped talisman, and this border pattern consists of alternating triangles in different colors. It is executed the same in all three pieces. One minor color change is noted in the BH piece. The colors used in the triangles in the S and RB pieces are red, blue and salmon throughout. In the BH piece only the inner border has this color combination. Its outer border is red, brown and salmon. Although commonly found in Saryk weavings, the tumar pattern is also found in Ersari and Arabachi rugs with some frequency. Rippon-Boswell's use of this border pattern as one of its justifications for a Saryk attribution is weak.

The use of the same pattern for the field and border is virtually without precedent in not only Turkoman but all other Central Asian weavings. The nearest design analog is a rare group of Beshir prayer rugs with a white field covered with floral or leaf patterns on interlacing vines. The same leaf/vine pattern is used for the single main border, but the ground color of the border is red, not white.2

Where the main border narrows across the top of the pieces, the kelle pattern changes to a plant form with three branches ending in the same fingered pattern as the upper pair of arms on the kelle. These alternate with blue and red stems in the S and RB pieces and red and brown in the BH. At the peak there is a red kotshak flanked by two red and blue diamonds on all of the asmalyks.

In comparing the color usage of the three pieces, only the RB asmalyk has a regular bottom left to upper right diagonal arrangment of a red-stemmed rosette kelles alternating with a blue-stemmed star kelles. In the S and BH asmalyks, the kelles are in a staggered arrangement with two left, then two right, then three left. The S and BH pieces end at the top with a centered kelle having a diamond top instead of a rosette or star. The RB asmalyk has a rosette kelle slightly off center to the right.

The number of kelles in each horizontal row is different in the BH and the S/RB pieces. From bottom to top, the BH asmalyk has 9, 10, 9, 10, 7, 4, 1 kelles per row or 50. The S and RB asmalyks have 10, 11, 10, 11, 7, 3 kelles per row or 52. All have 15 kelles in the main bottom border and three in the vertical borders. There are 11 plant forms on each side of the top border in the BH piece and 10 on each side in the S/RB pieces.

Along the sides and top of the field there are small gulyadi, chevron and small kelle patterns as filler elements to cover the empty space resulting from the offset number of kelles in each row. The drawing of the gulyadi in the RB piece is slightly different from the other two. The S asmalyk has a star and two boxes in addition to these patterns.

In Turkoman rugs the symmetric knot is found in some Yomud and Saryk weavings. It is also used by other Central Asian rug weavers such as the Uzbeks, Kirghiz, and Karakalpaks. The left-opening asymmetric knot has been published on pieces attributed to the Salor, Arabachi, Eagle Group, and Ersari as well as the Uzbeks and Kirghiz.3 Although certain knots predominate in weavings attributed to specific tribes, e.g. Salor and some Eagle Group 1 with left-opening asymmetric knots, this is not exclusive. Pieces with right-opening knots are known and published for both groups. The use of a specific knot type is not a certain and exclusive identifying factor for all the weavings of a group.

With specific reference to the current Saryk attribution of these pieces, based mainly on the symmetric knot, J. Thompson says in Turkmen, a comment on Pl. 4 in Bogolyubov and Sotheby's auction catalog of December 16, 1993, that the asymmetric knot was sometimes used by the Saryk and by the end of the century most ensis were woven with a right-opening asymmetric knot (Bogolyubov). Although published technical information on Saryk rugs does not confirm the use of asymmetric knots, I have noted it personally and know it to be true. Thus the presence of an asymmetric need not rule out a Saryk attribution.

A technical feature recorded on most Saryk weavings is a slight depression of the warps. There is no warp depression in the S and BH asmalyks. There is little or no warp depression in the oldest Saryk main carpets, so the absence of warp depression in these pieces may not be a valid indicator. But it seems unlikely that wedding pieces would have been woven with less than the best skill the weaver possessed, and warp depression requires greater skill on the weaver's part. Warp depression may not have been used in main carpets because they would have to be folded for moving, whereas bags were not. Folding requires more flexibility, whereas warp depression makes textiles less so. Most smaller items such as juvals and torbas were woven with depressed warps.

Scattered random patterns is not particularly common in most Turkoman weavings, but they appear to have been used even less by the Saryks. If these pieces are Saryk, that is certainly a deviation from tradition. Even Yomud asmalyks, which are the most numerous, have small random patterns mainly in the top portions where unusual patterns such as wedding caravans and jewelry patterns may be used along with scattered small filler patterns. Similar patterns sometime occur in the Tekke bird and animal/tree asmalyks.

With respect to the age of the pieces, based on the dye analysis, the BH piece could not have been woven before 1878-80. The simplest way to determine if the other two are contemporary with it is by dye analysis of at least the red dye. If it is synthetic, it is likely that all were woven at the same time. If their reds are natural, they may be as old as dated. Only the present owners could supply that data.

Is the BH piece copied from one of the other two? If so, it seems likely that the S piece provided the model because of the broken diagonal use of color. On the other hand, if a weaver were copying, would she not have copied the same number of kelles per row or the same random patterns at the sides of the field? These would serve as a guide to how many warps were needed to complete the pattern in each row. The weaver would not have to work out the variables and placement of the flowers.

On the other hand, given the very good condition of the S piece, why would the weaver feel the need to make a new one? It was still in good enough condition to use. Or were asmalyks only used once for a wedding and never again? Were they not handed down from mother to daughter or used for second or third daughters weddings? Turkoman women do weave new pieces when old ones become worn, but not until then. More information about the actual reuse of wedding paraphernalia would be helpful.

Or was the BH piece the prototype? Close observation of the flowers indicates that those of the BH piece are more clumsily and less consistently drawn than those of the S and BH pieces. For instance, the rosette flowers in the main border have three well drawn projections on each side of the flowerhead. This is not true in the field where the number of projections vary considerably as does the rounding of the top of the flower. All of the patterns are more crowded than in the S and RB asmalyks. If the S piece were woven second, the weaver would have had better control of the pattern with all the flowers being drawn the same. In order to gain more space, she may have eliminated one row of flowers and increased the number of flowers in the first four rows.

If the BH piece were the prototype, would the weaver have changed knot types? The propensity to use a certain knot type by some weavers has not been sufficiently studied and is not adequately discussed in the literature. There are enough pieces which deviate from the norm to know one knot type was not always used exclusively by the weavers of a tribe. We know that the symmetric knot was known and used by weavers who typically wove with asymmetric knots. We know this from the tentbands which are woven with symmetric knots over skipped warps. Thus weavers must have shifted back and forth in their weavings between the two types of knots.

Another feature of pattern adjustment occurs in the white at the center of the star flowers. In the main border and first field row, the white center is a diamond of nine knots. Beginning in the second field row, it is made of five knots. The five-knot center is used in both the S and RB pieces.

Just as the absence of a dye analysis from the other pieces inhibit an accurate comparison, so does the absence of full technical information. Particularly important is the type and insertion of the weft threads. If one or both of the other pieces have brown and white wefts, a relationship between the pieces would seem obvious. Absent that information one might ask why a white weft was used, when a brown weft is the normal practice by most Central Asian weavers. The answer may be that, since most white asmalyks have white kilim finishes to match the field, a white weft was needed to weave that area in white. Having spun white wefts for this purpose, she may have used white wefts with the more standard brown wefts in the pile.

Factors inconsistent with a Saryk attribution are the knot density and resultant handle. The pieces are more coarsely woven than most published Saryk pieces for which this information is given. The handle of Saryk bags and even main carpets is tight and firm; these pieces feel loose and floppy. The absence of depressed warps is another feature one expects Saryk weavings to possess. Although edge finishes can vary considerably in any group, the edges of most Saryk bags are a selvedge, not an overcast as in these pieces. All of these features are inconsistent with a Saryk attribution.

Another seven-sided asmalyk, attributed as Beshir, is published in Landreau, Yörük, Pl. 78. Although it has a loose weave and knot density of only 70 knots per square inch, there are no similarities in color, pattern or design concept. All of these Ersari pieces have the same characteristics one associates with Ersari weaving in general.

There are many rugs which have been attributed to the Ersari. As we learn more about all Central Asian weavings and become familiar with the Russian literature on them, it is questionable whether many pieces attributed to Turkoman weavers were woven by them or some other Central Asian peoples. Moshkova and Tsareva refer to some pieces as Uzbek Turkoman. Parsons in The Carpets of Afghanistan refers to the close societal and intermarriage of Turkomans with Uzbeks and their rug weaving. The early investigators of Central Asian weaving such as Dudin, Felkersam, and Bogolyubov all collected and attributed rugs to other tribes. Some rugs previously attributed to the Turkoman are being labeled Uzbek or Kirghiz by some dealers and specialists today. Elmby's attribution of the group of white ground prayer rugs to the Kirghiz is a case in point. In most cases, many of these rugs have Turkoman patterns, but the patterns are not as well and accurately drawn as one expects in Turkoman weavings. They tend to be more coarsely woven and frequently have brown warps as well as wefts. The palette and color usage is likewise slightly askew from Turkoman weaving.

Although no asmalyks have been attributed to any of the other Central Asian tribes, there is no evidence to show that they may not have woven them from time to time. They wove the same type of utilitarian items as the Turkoman, e.g. tentbands, ok bash, juvals, torbas. They also wove rugs such as julkhirs, which the Turkoman did not weave. Why might they not have occasionally woven asmalyks as well?

Although a sufficiently large data base on the weaving characteristics of these other tribes does not presently exist, from the data available it is evident that they used all of the same knot types as the Turkoman. Since they did not weave as many pieces, it may be that there was less consistency in the type of knot used by individual weavers. They too used the symmetric knot in their tentbands and would have been familiar with it.

In considering these three study pieces and their characteristics, there is not sufficient data to draw any firm conclusions as to their tribal attribution or age. In the absence of any clear and compelling evidence for a Saryk attribution, it would seem best to simply call these pieces Central Asian until more technical information is available on more closely related weavings. It seems to me that these could as easily be Uzbek or Kirghiz as Saryk.

There are numerous instances, primarily in dealer generated literature, about the tribe, age, and design characteristics of small groups of weavings such as these. Frequently these small groups of pieces are asmalyks. Two cases in point are the Yomud asmalyks with archaic field patterns and either weddings caravans or jewelry at the top and the Yomud asmalyks with five tree or plant forms on white or red fields. Although all of these pieces are attributed to the Yomud Turkoman, their dating in the literature varies considerably. In the case of the tree/plant asmalyks, technical data published on two of them show brown wool wefts in one (Tsareva, Pl. 74) and cotton wefts in another (Turkmen, Pl. 77). They may have been woven by the same tribe, but this is a significant technical difference.

Why is this important? These pieces are scarce and are thought by many researchers to be important to our understanding of pattern development and evolution in Turkoman weaving, if they are actually as early as thought by the authors. They also bring the highest prices in the market from collectors who, one would assume, want some assurance as to reliability of age and attribution. With many technical means available today such as dye and fiber analysis, a more scientific approach should be taken in evaluating such rare pieces when so few comparable companions exist with which to compare them.

References:

NOTES

2 See Tsareva, Rugs and Carpets from Central Asia, Pl. 98, 1984; Stanzer, Antique Oriental Rugs from Austrian Collections, Pl. 120, 1986; and Elmby, Antique Turkmen Carpets, Pl. 45, 1990. The Tsareva rug was first published by Felkersam in 1914 as "Uzbek-Beshir" and as Ersari by Tsareva. The Stanzer rug was first published as Spuhler/Konig/Volkmann, Old Eastern Carpets, Pl. 98, as Beshir; by Stanzer as Beshir Region; by Jourdan, Orientteppiche, Pl. 298, 1989, as Ersari Group; and Herrmann, ATT Band 3, Pl. 59, 1991 as Ersari Group. Elmby calls his piece Kirghiz(?).
3 See the technical analyses in Tsareva.