Daniel Walker: “Classical Fragments”

On April 19, 2008, Daniel Walker, Director of The Textile Museum, here in Washington, DC,

gave a Rug and Textile Appreciation morning on “Classical Rug Fragments,” featuring some fragments from his personal collection, as well as some others from those of Harold Keshishian and Wendel Swan.

What Walker described, when he began, as a visibly “hard-core rug morning” crowd filed in.

Harold Keshishian was among those who arrived early and said that the fragment of an Indian rug he is holding up is a favorite of his.

Dan started by talking a little about how he became interested in rugs and in fragments in particular.

He said that he was drafted into military service during the Vietnam war and was posted to the language school at Monterey, California. He was later stationed in Iran and first became interested in oriental rugs while there. He mentioned an Armenian dealer as influential in this early acquaintance. Subsequently, he took additional schooling in Islamic art history and became more seriously interested in rugs, in part, because nothing much was being done with them in the scholarly world.

He cited Charlie Ellis as a mentor, and said that he traveled with Ellis on occasion. Ellis was interested in rug fragments and Walker said that, likely, was how he first became infected himself.

Walker said that he has always been more focused on classical (pre-1800) Islamic material, and fragments of rugs from that era were all that he could afford. He also argued, as he did in some detail in his “Pieces of a Puzzle” exhibition on classical Persian weavings from Khurasan,


that a great deal can be learned from examining fragments and that collectors could, generally, learn a great deal by examining the pieces they have, regardless of condition.

Walker said that he would move in his presentation from west to east, but that there would be a noticeable emphasis on eastern material.

He asked Wendel Swan to talk first about two of Wendel’s Anatolian fragments.

Wendel has given me a copy of the notes he used in his portion of this program and I am using them nearly literally below.

Wendel said that, as he studied his fragments, his questions have been:

1. Where were they from?
2. What did the textile from which they came look like originally?
3. How old are they?

He started with the pile rug fragment below.


“…This is a fragment of a pile rug from Karapinar in the Konya province of Turkey. Trying to envision the full rug has been a real challenge. In the few months that I have owned, it I have found no “spot on” comparable.”

This fragment may descend from a classical tradition that includes a 17th century Karapinar rug in the TIEM in Istanbul. At least, the analogy is very close.

Here is the TIEM rug.

Using principals of symmetry, this sophisticated and well-drawn rug features reciprocal devices that are integral in Islamic art.

Note the less dominant cruciform device in the detail above and the lavish use of purple, a characteristic of Karapinar rugs thought to be from the 16th and 17th centuries. Also note the carefully articulated squinches (ed. corner brackets) in the corners, as well as in each half, of the field of hooked diamonds, the latter, also seen in my fragment.

At some point, the cruciform device became a mainstay of the Karapinar repertoire, commonly used in rustic rugs.

Occasionally the cruciforms were offset.

By the 19th century, many of these Karapinar rugs had lost the aspects of reciprocity, even though the field in my fragment might once have looked something like this.

The 19th century yastik below hints that so-called negative space may have been important in earlier Karapinar cruciform rugs.

The fantasy image, below, may suggest how the carpet originally appeared, but too much of the gold cruciforms are missing to know.

And an additional fantasy image below indicates how a more complex field design might have looked. But without comparables, this is pure speculation.

Here is a side-by-side comparison.

Wendel ended by saying: “For the time being, I can only believe that my fragment came from a rug that is between these two, both chronologically and from the standpoint of design evolution.”

Wendel’s second fragment was the one below. Again, the words that follow are mostly taken from his shared presentation notes.

This striped kilim fragment is complete in width, but the top and bottom ends are missing. From the wool and the colors, I knew it was probably Anatolian, likely central because of its colors.  Its palette is close to that from the Konya region.

However, stripes are found somewhere on most Anatolian kilims. Sometimes they are in between elements, as we see in the piece below.

And on the saf below.

Sometimes we see only stripes.

Here is a detail of the striping in the piece above.

Sometimes the stripes are on the ends.

However, the dimensions, the stripe patterns, and the colors, never quite matched.

When Jurg Rageth (author of a book on carbon-dating kilims) was at Wendel’s house a year or two ago, Wendel asked him about his kilim fragment. Rageth said that because of its traditional colors, he believed it to be between 150 and 300 years old, but that almost any date could be guessed.

Wendel reported that Rageth also said that he had no idea what the complete piece, of which Wendel’s kilim fragment was a part, might have been.

Wendel said that Rageth’s indications were “from the horse’s mouth,” so to speak, given Rageth’s experience with and knowledge of kilims. Wendel has subsequently continued his exploration of this “what the original complete piece likely looked like” question, and thinks he found the answer “in a camel’s mouth,” or more accurately on the back of the camel in the image below.

What Wendel saw in this photo doesn’t jump at you, but it’s there. This is a camel loaded for migration. Among the items tied onto its back is an Anatolian chuval.

The chuval is highlighted in color in the above version of this photo. It would be easy to miss it as a candidate for the format from which Wendel’s kilim fragment may have come, because the most visible part of it, in this photo, is brocaded with a design that features diamonds rather than stripes. But notice that there is a striped part of this bag, and that is what drew Wendel’s eye and tripped his memory.

“I remembered,” he said, “a similar sack in the Josephine Powell collection shown at the ICOC XI in Istanbul about a year ago.”

There are two brocaded panels on the front of this sort of Anatolian chuval, but the balance of such chuvals is composed of two striped sections that move around the back and are sewn together there.

Wendel: “So the sack would have looked like this when it came off the loom.”

“And like this, folded in half.”

“The brocading might have been cut off and used for other purposes, leaving use with this…”

“…which is quite a bit darker in palette than my fragment, but the scale and the number of stripes are quite similar…”

“…Perhaps most telling, the width of the Powell chuval is almost exactly the same as that of my fragment, 137 centimeters.”

|———————-137 CM ———————–|

For me (this is John Howe speaking now), this is a pretty telling demonstration of the probable source of Wendel’s nice kilim fragment. It seems very likely to have been one of two back portions of an Anatolian chuval.

Dan Walker now began to treat his pieces and Harold’s.

He started with the colorful Anatolian kilim fragment below.

Dan presented this piece oriented as it would be if it were hanging in an Anatolian dwelling. Many kilim books present their material rotated 90 degrees from this position.

Here is an image of this same piece in that orientation.

Walker said that the field design of this piece is sometimes described as a “double-niche medallion.” It is a fragment of one half of a kilim that would have been composed of two pieces sewn together on the sides opposite the border.

Dan mentioned that a device in the field design in this kilim is one implicated in the “goddess of Anatolia” interpretation and controversy. The device below is one version of what is sometimes said to be such a “goddess” image.

He called attention to the striping in these devices.

A great deal of Islamic art is geometric rather than representational, but Walker said that one of the other devices in this kilim is of an animal form. The detail below contains three such animals.

The top one is green, the middle one, red, and the bottom one is purple. They have seeming heads on both ends, reversed with horns or ears and an eye. Walker said he is confident that these are animal forms.

A kilim in Petsopoulos’ big book has similarities with Walker’s piece. It is his Plate 135, and is presented in his Konya section. No date is given, but he calls it “one of the most beautiful kilims.” The palette features a positive use of blue on a white ground, although the palette of the piece below is wider and brighter. The designs on this kilim are more dense and complex than are those on Walker’s example.

Here the ramshorns are less massive, but striping is more extensive. The side border is almost incidental and can almost be read as an instance of negative imagery. The animal forms include featuring on their bodies (skeletal?).

Walker suggested that his Anatolian kilim fragment was likely woven earlier than the second example above because its design is more spacious. 

Walker had a second, more austere, Anatolian kilim fragment. The one below.

What we can see are a narrow color palette and some delicate drawing.

He made comparisons with a kilim from the Vaklifar Museum collection. Walker explained that “Vakliflar” means “religious endowments” and that the kilims in this collection are identified by the mosque to which they were donated.

The piece above, Walker described as a more complete instance of a kilim with a field design similar to his red-blue example. Both of these first two pieces look quite old, in part, because of their simple, spacious designs.

A second Vakliflar example of this general type is the one below.

This kilim appears as Plate 16 in the Balpinar-Kirsch volume on Vakliflar kilims. The caption says that “This kilim is typical of tribal group segments from the region of Eskisehir, Afyon and Kutahya.”

Here is a closer look at a detail of this piece.

The age assigned to this densely patterned kilim is not clear to me. The record indicates that it “Collected in Sivrihisar (Eskisehir) from the Ulu Mosque (13th century).” I think the 13th century reference is to the age of the mosque.

I am not sure that Walker gave estimated ages for these three pieces. For me, both Walker’s piece and the smaller ivory ground Vakliflar kilim seem earlier than this third example.

I was curious about Walker’s use of “classical” in his program title. To some, the “classical period” of the oriental carpet might begin in the 15th century, peak in the 16th century and 17th centuries, and decline during the 18th century. He said that he had tried to limit himself to pieces estimated to be pre-1800 but that this next fragment likely spilled over that time line and he decided to make an exception in this case because it was such a tantalizing bit of a much larger pattern. It is the pile piece below.

Walker described this piece as Caucasian. He said the weave is very fine and the wool of a high quality and that it is, possibly, Shirvan.

He suggested that the field design in this fragment has on it part of a couple of “Lesghi stars.”

Peter Stone has done a study of various tribal devices and includes the Lesghi star as one of the Caucasian designs he treats. We should note at the outset that the Lesghi star device was used widely by weavers in many parts of the Caucasus.

Here is an image of a comparable section of a Lesghi star design from Peter Stone’s study of rug designs.

Seem similar? Notice that the “hooked diamond” filler device that occurs a couple of times in the Walker fragment is also present in this detail of the Stone image.

Now, here is the complete version of this Stone design.

Stone’s images are all computer-generated. The one above is based on a rug attributed to Daghestan.

And here is an image of a real Caucasian rug with a Lesghi star field design.  This rug is in The Textile Museum collection and was included in a recent TM exhibition on Mr. Myers’ collecting tendencies. 

Its significance is that it is a very early purchase (1905) by Myers and demonstrates that Myers started collecting much like many of us, with fairly ordinary Caucasian material that would likely not have attracted his attention later in his collecting career.

By the way, Walker says the weave of his fragment is fine. Stone examined a lot of Caucasians and noted that Kuba rugs are among the finest, averaging 110 kpsi. Walker’s fragment could well be Shirvan, but if it has depressed warps, it might also be Kuba.  (In his recent catalog for the 75th anniversary of the NYC Hajji Baba Society, Jon Thompson indicates that there is no reliable way to distinguish Shirvan pile rugs from those attributed to Kuba.)

Walker next moved to Persia with the fragment below.

The design has red flowers on a blue ground with a medallion, striped on its edge, in the upper right hand corner.

Walker said that most see this fragment as having been woven in NW Iran or by Kurds. It has a cotton foundation and the wool pile is of very good quality. He made analogies with a rug from the Metropolitan Museum of Art.

The design of the Met rug features rows of cypress and blossoming trees growing out of stylized pools of water. The devices are staggered row to row so as to form a kind of vague lattice. Here is a closer detail of this design.

Although Walker’s fragment here may not have been stepped like the Met rug, it seems possible that it may have been part of such a design. And, although the distances seem different, the “medallion” in the upper right corner of the fragment might well be one of the pool representations in a row above. This image is from Arthur Pope’s “Survey of Persian Art.”

Walker indicated that this fragment was published in F. R. Martin’s famous book on oriental rugs and may well have been owned by Martin himself, since the image lacks a credit line. In this context, he noted that Chapter 8 of Martin’s book was reproduced with black and white versions of its images in Oriental Rug Review, Volume 5, No. 12, March, 1986, pp. 11/547-17/553. Some other designs similar to this fragment are treated in Chapter 8.

The next set of fragments is attributable to Kerman. Walker said that there is a group of rugs often described as having been made using the so-called “vase technique,” a name meant to describe a particular weave structure but taken from the appearance of vases in some, but not all, of the carpets in this class.

Here are the three Kerman fragments Walker showed that were made with this “vase technique.”

“Vase technique” Kerman fragment 1

“Vase technique” Kerman Fragment 2

“Vase technique” Kerman fragment 3

The expression “vase technique” refers to a distinctive structure used to make these Kerman rugs. There are three wefts between each two rows of knots. Two of these wefts (numbers 1 and 3), usually wool, are inserted, taut, straight through separate sheds. The middle weft is silk and is inserted in “wound” fashion. The drawing below of this “vase technique” structure is taken from an essay by Jon Thompson in “The Persian Carpet,” an auction catalogue published by Lefevre & Partners, London, 1977, p. 71.

Walker said that this structure is faulty because, when the pile gets very worn, the wound silk wefts tend to break from exposure to foot traffic.

Here is a detail of a damaged part of one of the “Vase technique” Kerman fragments above. The places where the white cotton warps are visible are among those where the silk wefts are damaged or broken. One can also see the wool wefts where warps are broken.

Someone from the audience asked why the Kerman weavers would choose a faulty structure. I don’t think Walker answered directly, but one could conjecture that it may have had to do with the desire to have a “compact” weave.  Given the mostly heavy cotton warps in these rugs, fine silk, which is known to have great tensile strength, may have been considered ideal for pulling this pile structure together compactly.  These Kerman weavers may not have considered that silk does not have great tolerance for abrasion and that it might break with wear.  In a subsequent conversation, Walker, gave another plausible reason for this latter possibility.  He said that he thinks that most weavers, including those who wove the “vase technique” rugs, weave with rather short-term use objectives and don’t really consider what might happen to a given structure over longer spans of time. In this particular case, the structural weakness is not apparent, and in fact does not occur, until the pile has worn down almost completely. 

There is one additional structural feature to notice about some of these “vase technique” Kerman rugs. A few carpets preserved only in the form of fragments have pile on the back as well as the front.

Here, below is one of the “vase technique” fragments that has its shaggy back being turned so you can see it.

And here are two straight-on shots of the fuzzy backs on two of these fragments.

Kerman “fuzzy back” 1

Walker said that he was not entirely sure of the purpose of this extra pile, but the conventional explanations are for extra thickness or warmth, or to prevent slippage.

Walker noted that “vase technique” rugs were the focus of May Beattie’s book “The Carpets of Central Persia.” By concentrating on the distinctive “vase technique” structure, Beattie was able to demonstrate that a variety of pieces, previously thought to be members of quite different groups (primarily on the basis of design), might defensibly be grouped together. Beattie’s analysis and findings are one of the early impressive applications of structural differences to rug scholarship.

The fragment below is Plate 10 in Beattie’s book.

This piece is part of The Textile Museum collection. Beattie divides the various designs used in “vase technique” carpets into eight groups. The piece above is placed in her “three plane large leaf lattice” design group. Based on its similar pattern and scale and also on the orange wool shag present on the back, Walker speculates that his three fragments come from the same carpet as the large TM fragment or perhaps from a similar carpet. Wendel Swan pointed out, from the audience, that there is a second Kerman fragment with a shaggy back in the TM collection.

Walker moved next to five smaller fragments of his which he thinks may be attributed to 17th century Kashan. They are quite scrappy in appearance but in fact have a lot to tell us. 

Here is the long fragment from this set.

A closer look at the piece above.

The other four fragments, all from the left border of the rug (going by pile direction) are presented in the following two images.  There are two fragments in each of them, joined along the center line. Note that one fragment in the first pair shows the lower left corner of the border.


These pieces are very finely woven and are entirely of silk.

Walker said that they are instances of fragments that tell us a great deal about the complete rugs from which they came. Although the four smaller fragments are from the border, the larger fifth piece shows enough of the field design to permit us to reconstruct the entire pattern. We can even speculate the likely size of the original piece from the scale of the pattern elements and the border. 

Walker said that these fragments are close to the 16th century silk rugs generally attributed to Kashan in terms of materials, weave, and palette, but they bear a pattern that is not seen until the middle of the 17th century. The best example of this pattern is a silk rug of another type – the famous Polonaise class – that belongs to Rosenborg Palace in Copenhagen. The Coronation Carpet, as it is known, was brought to Europe in 1664 by the Dutch East India Company. Here is a large detail of that rug.

The image below is of an even smaller detail of the field of immediately preceding piece.

The field pattern here consists of a rather tight cluster of scrolling vines , blossoms, leaves, and cloudbands forming a pattern unit that is repeated many times. This is the same convention followed in Walker’s Kashan field fragment.

No other 17th century Kashan rugs are known. Moreover, he said that there is jufti knotting in his fragments. Since this is not a characteristic of 16th century Kashan rugs, it indicates to him that the quality of Kashan production was deteriorating somewhat by this time.

Next Walker moved to two old fragments that belong to Harold Keshishian. They are part of a Persian group sometimes described as having “vinescroll and palmette” patterns, usually on a red ground with dark blue or green borders. This group has had many names over the years – Shah ‘Abbas, Isfahan, Herat, and, more recently, Indo-Persian – and this suggests the uncertainty that has long surrounded the origin of these carpets.

Both May Beattie and Charles Grant Ellis believed, based mainly on issues of color, that certain examples of this type were woven in India. Although there are indeed North Indian carpets based on this pattern, they can be distinguished by structure and color usage from the Persian rugs.

Most scholars today believe that the vinescroll and palmette type is Persian in origin and dates from the 17th century.     

 Here are four images of the larger of Harold’s vinescroll and palmette fragments. It gives a good sense of what the border and field looked like, and the scale of the pattern elements suggests that this was once a fairly large carpet.

Closer full length below.

Detail of field below.

Detail of border below

Harold’s second vinescroll and palmette fragment shows only a portion of the field pattern.

Some of the reds in other of these large scale fragments would tempt one to suspect a lac dye, and the one above seems particularly tempting.  The presence of seeming lac reds might seem to make an Indian attribution more likely, but Walker said that lac was both available and used in Persian as well as Indian rugs and so the presence of it does not by itself permit this distinction.

Walker pointed out that vinescroll and palmette carpets were commercial products made in very large quantities. Well over two hundred examples survive today in collections in America, Europe, the Middle East, India, and Japan. They are represented in many Dutch paintings, particularly during the periods 1620 to 1630 and 1660 to 1680, when their popularity must have peaked.

He then showed two examples of complete vinescroll and palmette carpets in order to place Harold’s fragments in context.

The rug shown above is a well-preserved smaller example that belongs to the Boston Museum of Fine Arts.

Some examples of the type are extremely large, like the one that follows from the Jaipur collection. 

The above carpet measures almost 32 feet long and is one of a pair. Several other examples are even bigger.

The next piece belongs to Walker and is a fragment of an Indian rug.

It has pashmina pile on a silk foundation.  Walker said that it is early 18th century and was likely woven in northern India.  It has about 500 knots per square inch.  This is pretty fine but doesn’t begin to approach a few 17th century pashmina rugs whose knot count is 2,000 or even 2,500 per square inch.

Several other small fragments the same size and shape as Walker’s survive, in the Hermitage, the Boston Museum of Fine Arts, and two in private collections. Only one larger fragment of this rug is known, and a large detail of it is shown here.

The large fragment measures 9′ 7″ by 8′ 2″ and lacks most of its borders. It belongs to the Cleveland Museum of Art.
The Cleveland fragment and the smaller pieces that once adjoined it have a pattern that is called “millefleur,” after the many small blossoms that are arranged in clusters. A famous millefleur rug that survives in complete form is the 17th century example below, that features a niche in its field. It belongs to the Museum of Applied Arts in Vienna and is illustrated as Figure 128 in Walker’s “Flowers Underfoot” catalog, 1997, page 131.

The image below is a detail of a part of the field of the overall image above.

Next Walker turned to the Indian fragment of Harold’s that Harold held up Just before the beginning of the program.

Although it is a very small bit, it is nonetheless recognizable as from an Indian flower and lattice rug of the 17th century. These were made in different quality levels, from the commercial grade represented by Harold’s little piece, to the luxury grade seen in a couple of pashmina examples. The image below is of a closer detail of Harold’s piece.

Harold brought in a second little Indian fragment that he acquired together with the first, but Walker could not be sure whether it came from the same rug or not.

A closer look at part of it.

Walker, then showed an example of a complete flower and lattice rug, not of the pashima quality, but of a commercial grade close to that of Harold’s piece. It is also very close in design to the rug Harold’s fragment was part of.

Carpet (detail), India, Mughal Period, 17th century; Wool pile on cotton warp, silk and cotton weft; The Textile Museum 1994.12.1; Gift of James D. Burns

 Walker’s program came to an end, he answered questions and folks flocked to the front of the room to get closer looks at these classical fragments.

My thanks to Dan Walker for allowing me to share this interesting RTAM program with you. Dan has also, generously, “looked over my shoulder,” editorially, to make sure things are right.


R. John Howe


Comments are closed.

%d bloggers like this: