Christopher Alexander argues. in his book on Turkish village rugs, that a rug design cannot achieve full aesthetic merit if (among other things) the design does not take advantage of opportuities for variation that are available at the level of the individual knot.

And one of the most usual rules we recommend to one another, is that since, changing rug structure is potentially more costly to weavers than changing designs, structural aspects, like whether the knot in a piece is symmetrical or asymmetrical, tend to be more reliable indicators of where a rug was woven or by whom, than do the designs used in it.

But it seems to me that we often act as if we do not agree about these two "knot-level" rules.

For example, most folks feel that Talish rugs with an open field are aesthetically superior to those with designs in the field, a clear situation in which the opportunities for design variation at the level of the individual knot have been passed by.

But even more potentially perplexing for me are two instances I have encountered about the role the character of the pile knots, in two pieces I own, should play in their attribution.

The first instance is a Turkmen mafrash that I own, whose designs appear to be classic Yomut. And that is what I have thought it was for some time. But recently I've had the occasion to look closely at the knot (which I had never done, since the designs so obviously pointed in the direction of a Yomut attribution). To my surprise, the knots are asymmetric open to the left. Of course, that finding makes one wonder if one has an "eagle" group piece that one has not noticed before. (The wool is very high quality and the colors are similar to some eagle-group pieces.) But none of the other structural indicators seem present (e.g., traces of cotton or silk wefts). I have asked several experienced Turkmen collectors about this, saying has anyone seen a seemingly Yomud piece with asymmetric open left knots that is not an eagle-group piece. The answers have been disultory and unsatisfying. It seems to me that no one else treats this finding as as anomalous as it seems to me.

In the second instance, I have recently biought another piece, which the seller estimated to be Yomut. A friend of mine, far more experienced than I, suggested that the compartmented design of the piece suggested to him that the piece is likely Ersari. When I pointed out that the knot is symmetric (and that I know of no Ersari pieces with symmetric knots, excepting at the sides), he held to his design-based view, saying that "you Turkmen collectors treat knots more importantly sometimes that you should."

So my question is what do we think of these two knot-level conventions? Do we actually use them in ways that suggest that we believe them to be true or is our usage more opportunistic? That is, do we adopt them when they support an argument that we want to make on other grounds and largely ignore them when they conflict with assessments we make that are not based on them?

Regards,

R. John Howe

My own position is - and has been for a long time - that the criteria we use for attribution (like knot type) are not absolute. What they do for us is add weight to the probability that a piece is of a certain geographic origin and/or was made in a certain time. Thus, if most other criteria pointed to Yomud but knots were asymmetric open left, I'd go with the likelihood that it's Yomud.

As a related aside, I am quite confident that the motorcycle seat cover now on our Show and Tell board is Yomud, and it has asymmetric knots open to the left.

Regards,

Steve Price

Of course, the claim has never been that structural indicators are "absolute," but rather that they should weigh more heavily than design indicators, since there are good reasons for believing that weavers are more reluctant to change structural aspects of their weaving than to adopt different designs.

It sounds like you treat design and knot type indicators as nearly on a par with one another. True?

Regards,

R. John Howe

I should also acknowledge the usefulness of the information you gave on your bicycle seat cover.

You feel that it is very likely Yomud (and not an eagle group candidate although note that none of the eagle group pieces appear to have great age) and has an asymmetric knot open left.

Perhaps this suggests that this knot is a more recent Yomut usage.

Regards,

R. John Howe

I don't think we can make very specific generalities about the relative importance of each criterion. Design is clearly extremely important at some levels. A Kazak (to pull one example out of the air) has symmetric knots, and is clearly not Yomud. This would be a judgment made on the basis of design, palette and knot density. Saryks have symmetric knots, are obviously not Kazaks, and can be distinguished from Yomud work by the materials (cotton and silk, frequently), palette, and designs.

The motorcycle seat is, I think, a nice illustration of some worthwhile points. I don't think anyone would argue that it's Salor or Eagle group; the design and motifs are of Yomud tradition. Later pieces are more likely to deviate from traditional norms than earlier ones are, so perhaps we shouldn't be too surprised at the asymmetric knots opening to the left in this one. But deviations from the norm must have begun as fairly infrequent occurrences and become progressively more common (the alternative is that each deviation arose abruptly and became fairly common almost instantly).

Regards,

Steve Price

Regards, Yon

Why do you say tribal attribution of mid-20th century Turkmen stuff is meaningless? People like Jon Thompson and Peter Andrews were still doing field work among Turkmen until rather recently (they may still be doing it for all I know), and they refer to the tribal names of their subject groups. So the tribal identity still existed (exists). Do you think the weaving traditions have become so garbled as to lose their characteristics? Why? About when did this happen?

Regards,

Steve Price

Regards, Yon

I agree that much of the 20th century Turkmen production comes from Soviet-run workshops and has no tribal identity in the sense that you and I understand the term. Still, the items are easily attributable to Turkmen.

On the other hand, there's plenty of 20th century Turkmen stuff that is (or was) not made in workshops, but in the tribal community. I believe most 20th century bags are in this category, and I can hardly imagine that my motorcycle seat cover is anything except a utilitarian article produced within a tribal community for use within the group. If this is the case, it is not meaningless to ask which tribal community it was. The layout and motifs are Yomud, so I believe that it was made in a Yomud setting, in spite of the the asymmetric left knots.

Regards,

Steve Price

Steve talks about comparing Central Asian to Caucasian rugs above and I don't think most of us are worried about knot similarities there. Knots, as I understand it, (although this conversation is making me wonder about what I thought I had learned) are more usually used in closer comparisons. With Western Turkmen pieces, to help make tribal attributions.

Yon writes indicating that he has a piece with a knot that is asymmetric open left that seems both Yomut to him and likely not a eagle group member. He also points out that the Yomut group is a large one and hasn't been well sorted out so far.

I think my "Yomut" piece with an asymmetric knot open left is considerably older than Steve's bicycle seat but since a number of us lunch off design more than I think I do, here's a direct scan of perhaps one third of the pile side of this piece so that you get some notion of what it is like both with regard to design and color palette.

It seems me that Yon's indication here suggests that others are also sometimes encountering asymmetric knots open left on pieces that seem Yomut but not eagle group. So maybe we've done what we can with that piece.

No one has said anything about the notion that a piece might seem Ersari enough in its design to make us stick to that judgment despite, the possibility that the piece might have a symmetric knot.

Just so you have some design impressions, here is an overall image of this piece.

t looks Yomut to me but my experienced friend thinks it's likely Ersari because of the "tile" design. The dealer I bought this piece from said that the knot was symmetrical and that 's what I thought it was too when I looked at it on receipt but today at the TM folks said that the knot is asymmetric open right. This means that I have to go to school again on my knot recognition skills and that it is possible on the basis of the character of the knot that this piece is Ersari, although it could still be Yomut too.

(There are a couple of other things that are likely unrelated but that I should mention. The dark blue looked "royal-ish" enough on my monitor to make me wonder about it and the wefts in this piece are a quite brightish orange. I've seen brightish orange in Yomud pile but not as weft. My concerns about synthetic dyes were not shared much by the folks at the ICOC X LOC meeting who looked at this piece with me)

I invite other opinions. Do you think this piece is Yomut or Ersari? If the dyes are natural, would you propose a likely age for it? (Don't be shy. I'm prepared to discover that it's a 20th century piece. ) Reasoning?

Regards,

R. John Howe

Your "tile" design rug looks mighty Ersari to me, and if it has asymmetric knots opening to the right, there seems little reason to seriously consider alternative attributions.

I have no opinion on its age. Obviously, it's lived long enough to show some wear.

Regards,

Steve Price

While I'm eating humble pie this morning about my ability to read knots accurately, let me share a sentence just encountered from the Pinner and Eiland catalog on the Wiedersperg Collection.

I've been a fan of this catalog, despite the fact that Michael Franses has written quite critical things about it, because it seems to me to be an effort by Pinner and Eiland, not just to discuss this collection but to give some sense of the state of the art in Turkmen studies at the point that they wrote it.

Anyway, I started the day yesterday, pretty firmly believing that if a rug had symmetric knots it was highly unlikely to be Ersari, and in truth I have never seen an Ersari piece that uses them predominantly rather than at the sides to stabilize the last column of asymmetric knots on the side to which they are open.

But at the end of this Pinner-Eiland volume, Murray Eiland offers some "Notes on the Structural Analysis of Turkmen Rugs." In it, today, I encountered the following disconcerting sentence:

"...Occasionally one will find a rug that appears to be Ersari with symmetric knots, but usually these are types that could have been woven by Uzbek groups..."

So even if I had read the knot in this piece correctly (i.e., if they had been symmetric) there is a chance that my experienced friend's suggestion that it is Ersari (and this was based initially on design, and maybe color palette alone) could have been correct, despite that structure. It might have to fall in the more Uzbek-like portion of that group, but it would still be a possibility.

As Cecil Edwards, says in his great book in a slightly different context, "Thus do our fondest illusions die."

Regards,

R. John Howe

The equivalent saying in my racket is that that it is a sad moment when a beautiful hypothesis meets an ugly fact.

More to the point: it is very useful to remember that attribution criteria are hardly ever absolute. They simply lend weight to certain possibilities, and are implicitly statements of probability. The First Law of Statistics is that unusual things happen now and then.

Regards,

Steve Price

Regards, Yon

P.S. John, The Ersari-Uzbek attribution of your piece seems reasonable to me.

I cannot follow your knot analysis. However, from a purely design standpoint have you considered a Chodor tag? I have seen some Chodor main carpets that vaguely resemble this piece.

Good luck, michael

How's our old town?

To clear up the knot analysis confusion, the dealer and I read this knot as symmetric (the pile is very worn). Bob Emry and Wendel said it was clearly asymmetric open right. I have never seen an Ersari rug with predominantly symmetric knots. Most folks are saying that the designs suggest that this piece is Ersari. The asymmetric knot forces me to admit that possibility, although the border look very Yomut to me. Eiland says that even if the knot is symmetric, the piece could still be Ersari, Uzbek sub-group.

About your thought that the piece might be Chodor, there is no hint of cotton in the wefts. The wefts are a brightish orange wool. The warps have that "brush-like" separation of fibers and a kind of stiffness, that I've come to associate with goat hair.

Hope things go well there.

Regards,

R. John Howe

I'm afraid I don't see the border on your tile-design rug as uniquely Yomud. To my eye, the most Yomudish characteristic it has is the soldat motif in the bands seaprating the tiles. Chodor is plausible to me, but the design seems so Ersari, and there's so much more Ersari stuff around relative to Chodor, that my inclination is to default to that attribution in the absence of better indications to the contrary.

Regards,

Steve Price

Yes, the Yomuts use this "tuning fork" design a lot and I admit that my sense of attribution for this piece is (now absent a symmetric knot) largely speculative but let me ask another question of those who think it seem Ersari.

One of the reasons this very tired piece interested me was that I do not think I have seen this design before.

Wendel Swan thought he had actually seen another piece with a similar design that was designated as Ersari.

Since, despite a knot that allows us to think this piece is likely Ersari, we're seemingly lunching mostly off design.

Given that, what is the closest piece to this design that anyone has seen? Is there an image of it that we can put up?

Thanks to all,

R. John Howe

I've seen the tile design in, I believe, Skinner's catalogs and also in HALI, although I don't have specific references handy. The attributions were Ersari, which means that the pieces really were Ersari or that the people who did the attributions were wrong.

You wrote a couple of times in this thread, we're seemingly lunching mostly off design, as though that was odd, perhaps only marginally respectable. The fact is, almost all geographic attribution is based mainly on design, motifs and palette and we use structure mainly to fine tune within broader groups.

If you hadn't originally read the knots as symmetric, I doubt that Yomud would even have entered the picture.

Regards,

Steve Price

We may now be living out Jerry's main salon title at levels he did not really envision, but let me pursue a couple of things Steve said in this last post, in relation to what I have been told during my indoctrination into the rug world.

First Steve wrote in part:

"...You (ed. the reference is to me, John) wrote a couple of times in this thread, we're seemingly lunching mostly off design, as though that was odd, perhaps only marginally respectable. The fact is, almost all geographic attribution is based mainly on design, motifs and palette and we use structure mainly to fine tune within broader groups.

My thought:

While it is true that attribution analysis is not a single variable business: what we are looking for is to have several indicators point in similar directions so that we can move to conclusion with a degree of confidence, it has been argued to me, from almost my first day in rugs, that these indicators are not always of equal weight.

I started this thread by rehearsing one argument for treating structural indicators as having more weight than some others, including specifically rug designs. It is that it is markedly costly for weavers to change structural features (I think this may have been particularly true of 19th century tribal weavers). Much more costly than it is for them to weave a different design. This is the reason for believing that structure should usually be treated as having more weight in disputes about attribution than should given design features. Design features move quite quickly and easily, sometimes over great distances.

So while I agree that most of us treat designs as perhaps the most important indicator of geographic attribution, there is reason in this argument about the character of structural variation, to question whether this is appropriate. It seems to me that the appropriate move that we should make if we want to treat design on a par with structure is to show that the argument that structure is a more stable indicator than design is incorrect in some sense.

It does not help much to argue at the level of assertion.

Steve also said in part:

"...If you hadn't originally read the knots as symmetric, I doubt that Yomud would even have entered the picture."

Me:

Well, it may not have come through in my telling, but the first mention of a Yomut attribution was by the dealer who sold me this piece. So although my misreading of the knot contributed to the confusion, the Yomut attribution was there in my own introduction to this piece. My first question was what is the knot and the dealer's answer was "symmetric."

I do not want to seem to protest too much about the seeming consensus that this piece is Ersari (and that may be absolutely correct) but I would be interested in references to or images of other similar pieces.

But Steve is accurately reading my understanding of the relative importance with which we should treat structural indicators vs. those related to design and pattern. Not as a basis for "knock-down decisions," but as having discernibly more weight than do design features for good reason.

The good reason might be successfully attacked but, as far as I can tell, that remains to be done.

Regards,

R. John Howe

Please Click on "Next Page" (below, to the right)for the Continuation of this Thread

I believe that the weight to be given to structural considerations vs. design, motifs and palette depends on the piece and the specific structural element we're talking about.

In your tile design piece, if it actually had symmetric knots I'd think it a tough call between Ersari and Yomud. The two sets of criteria would contradict each other, and other than Chodor I don't know of any other sensible alternatives that would resolve the conflict. Since it has asymmetric knots open to the right (rare in Yomud, common in Ersari), and the design looks Ersari, I think the probability that it's Ersari if rather high.

Here's another example, taken from a previous Salon.

The image isn't very good. The piece has Salor color and motifs, and asymmetric knots open to the right (as do 20% of Salor pieces, according to Jon Thompson). But the world believes that the Salor made this design only with 3 guls, the 6 gul version being introduced by the Tekke not too long after nearly exterminating the Salor. Thus, on the basis of structure it's more likely Tekke but Salor is still in the running; palette and materials give more or less equal weight to each; design makes it unambiguously Tekke (if the world's belief about the number of guls is correct).

Regards,

Steve Price

There are many examples of tile pattern rugs attributed to the Ersari in the literature. Some have hour glass designs within the tiles, some have diamond shapes, while others are geometricized flowers. A few years ago I considered acquiring another type with a serrated diagonal line bifurcating two colors.

Not long ago we had a discussion about the complex process by which we recognize and label rugs. Sometimes that process is inconsistent and at other times it is inaccurate. When I saw the image of the rug John now owns and brought it to his attention, all of my senses told me (without examining it) that it is Ersari. Others might call it Kizil Ayak, but I wouldn't have thought Yomud or Tekke or any other Turkmen group.

Following is plate 88 from Old Oriental Rugs by Spuhler, Konig and Volkmann, which they call Ersari, early 19th Century.

It has the "tuning fork" borders as does John's. It has asymmetrical knots open right and very light gray warps, just as does John's. The Spuhler piece, however, has brown to dark brown wefts. It seems to be fragmentary.

While the designs within the tiles are dissimilar, there is no doubt in my mind about the connection of this rug to John's. They may not have been woven by exactly the same Ersari group, but I suspect no one will ever know the precise answer. I'm not sure that anyone knows enough to make a finer distinction.

Best to all,

Wendel

Following is the image of the rug I considered purchasing a few years ago. While I continue to admire it, it just didn't fit into my collecting pattern and I was able to see it again earlier this year in its New England home. Hali referred to it as Ersari (Kizil Ayak?), each probably as meaningful or meaningless as the other, but I think we still have some sense of what these terms convey. The tiles in this rug are separated by borders with the tuning fork device.

Wendel

My first thought seeing the picture of John's rug was "Tekke". I had the impression of a rather small, densely knotted rug, reminding me to a Tekke in the style of Jourdan's No. 50, 51 and 53 which has a rather big share of white in its palette too. In my eyes also the very disciplined way of colour arrangement in two sets of interwoven diagonals spoke for Tekke.
But 'unfortunately' I came in "Alte und antike Knüpfarbeiten der Turkmenen", Bernheimer, Munich, 1977 across the following "Ersari-Beschir". The scan shows about a third of the carpet measuring impressive 290 x 626 cm.

And your carpet, John, isn't small neither - is it?
The two topmost rows of compartments show the same design as John's carpet does. And, although only with difficulty to see on the scan, there is the same system of diagonals for the colour arrangement. (My association of 'discipline' with Tekke was obviously wrong...)
Therefore I now would opt for Ersari as well. But having in mind what already Wendel indicated above: Not only (structure and design) analyses have their limitations, the labels we try to derive from them have the same limitations too...

Best regards,
Christoph

Wendel and I have exchanged some messages on the side as this discussion has gone along.

This morning, before I saw either of the images he has provided above or that provided by Christoph Huber, I wrote Wendel saying that I had taken a quick tour of some of my Turkmen rug books last night and that the closest rug I could find to mine was one in Hans Konig's article toward the back of the Macke/Thompson catalog. That photo is a black and white image of the first piece that Wendel has posted in color above.

Konig says that this piece is an example of second of four Ersari design groups that he identifies. Here is his description of this group:

"...The second design or quasi-tribal group includes large main carpets which are Turkman in feeling, color and design but whose patterns do not conform to the criteria usually accepted for tribal main carpets. Guls are either entirely absent or are arranged differently from those in Tekke or Saryk rugs, for example. Diamond forms, lattic patterns, and compartments, (Figure 58 [ed., the first of Wendel's images]) occur often in this group, reminding the observer of certain Yomut or Chodor carpets with similar design composition..."

This last sentence has a footnote which reads:

"The execution of this compositional scheme and the resulting visual impression are, of course, in the case of the Yomut and Chodor carpets, different from those of the Ersari carpets."

It's a bit infuriating that Konig hasn't taken the next step here and either said what these differences are or moved to give us an example or two.

There are a couple of additional things that might be useful to note.

First, Wendel has had both the pieces he posted and mine, in his hands, so he knew the actual colors, handle and weave patterns. As you can see on this site, I've explored a bit how we work with both handle and weave pattern and it seems to me that mostly what goes on is that experienced collectors remember basic similarities or differences with regard to both of these dimensions and that these are retained and applied on a recognition basis by them often with quite remarkable accuracy (very much like our ability to recognize faces).

And, of course, Christoph's example seems to function as a clincher, since it provides the first instance I have seen of the very same pattern. (But notice that Konig allows that nearly everyone in the discussion could be right.)

I AM in Wendel's debt about owning this rug, but not in quite the way he has suggested above. I had seen this piece, without his cue, when it first appeared, had printed off the image and had asked some folks a bit about it. (but by happen stance, not Wendel, yet) Then the piece disappeared for awhile. When it appeared again, Wendel did indeed flag it for me. But the real favor he did was to say (in error) that the seller was asking less than half of what he in fact was. This galvanized me to inquire. The dealer was at first astounded at the price I had heard and said that it was "impossible," but then when I made an argument suggesting that although the piece was "interesting" its condition and questionable dyes made it likely of "marginal collecting interest," my offer was almost unbelievably accepted.

To paraphrase the Scriptures, somewhere: Greater love hath no man than that through which he makes it possible for a friend to buy an interesting piece at less than half price."

Thanks, Wendel, and to all of you for this nice discussion.

Regards,

R. John Howe

Postscripts:

Christoph, my piece is about 4 1/2 feet by 7 feet.

The two pieces that Wendel posted and, I think, Christoph's too, seem to have similar color palettes, although the proportion of red in Christoph's example is quite different.

The first of the tho rugs Wendel has posted here has both the white and yellow ground panels that appear in mine (but mine also has some panels with dark blue and red grounds).

Wendle mentioned what is sometimes described as the "Kizil Ayak" subgroup of the Ersaris and says that we likely know what we mean when we indulge in that usage. My current understanding is that it is mostly a distinction based on fine-ness. Most Ersari main carpets tend to have about 50-70 knots per square inch but can reach 100 knots per square inch and the bags can often be higher than that. Kizil Ayak main carpets seem to be finer, starting at about 100 knots per square inch.

I haven't counted but one thing that Wendel noticed was that the weave in my rug appears to be quite fine.

I have also referred to the weave pattern in my piece. This is something Wendel noticed that also seemed different to him than Yomut pieces he had seen. The wefts in my piece are placed so they seem always at a slight angle as they move across the rug. Not as wavy as some Shirvans or Sennahs but something like that. I don't have my only instance of a seeming Yomud piece with an asymmetric open right knot at hand, but my memory of its weave pattern does not include this slanting of the wefts. So I think Wendel is likely right about the weave pattern in this piece being distinctive.

Regards,

R. John Howe

I should have checked one book more: Jourdan, Orientteppiche (Augsburg, 1996, not to be confused with the volume about Turkmen carpets) No 385 (also advertised in HALI 55 p. 17). This carpet has almost the same design as John’s, but has the colours of the compartments arranged only in simple diagonals. Just as the Berheimer piece this rug is very large ( 347 x 590 cm) and Jourdan speaks of a distinct nearness to Yomut pieces...

In relation to his second design group König writes in the above mentioned paper: “[...] I do not believe that alternate layouts result from truly authentic Turkmen patterns. More likely, several design principles coexisted in Turkmen tradition.”
Though it would be perhaps necessary to discuss what “truly authentic Turkmen” means, I fully agree and I guess that the design principle we are dealing with here is derived from flat woven textiles. The little flowers in the compartments of the first Ersari above are strongly reminding the (brocaded) ones on Turkmen tent bands and the overall impression (apart from the colours of course) of these compartment rugs is not unlike to one of Vernehs such as Eder, Kaukasische Teppiche No 459.
This hypothesis would pose several new questions, some of them difficult to answer with the rather poor knowledge about Turkmen flat weaves we have. But it may could explain why this kind of designs seems to occur in several/all major tribes and this without the urge to resort to intertribal marriages and the like.

Best regards,
Christoph