Regards, Yon

It’s a straw man, alright, but it may not be so easy to knock down.

The juvals we call Tekke today (from their motifs and palettes, I hasten to add) are essentially all knotted asymmetric open right, while the ones we call Salor are (according to Jon Thompson) asymmetric knotted open to the left about 80% of the time, about 20% are knotted asymmetric open right (just like the ones we call Tekke!). The distinctions that Thompson draws between Tekke and Salor (and the distinctions that we draw) are almost entirely design/palette/motif based.

The turreted gul juvals are divided between the two tribal attributions mostly on the basis of design. The six-gul specimens are just about always attributed to the Tekke, and I believe that every known example has the typical Tekke knot (opened to the right). The three-gul specimens are attributed to the Tekke if they’re knotted open right, Salor if they’re knotted open left. But, of course, 20% of the things we call Salor are knotted open right, so the Tekke attribution of a three-gul juval that’s open right has a 20% chance of actually being Salor. That’s not a knockout punch level of attribution confidence, in my opinion.

How about eagle gul group? They do, indeed, have some structural characteristics that take them out of the Yomud mainstream, although the foundation on which the group rests is still - guess what? - the eagle gul.

Regards,

Steve Price

As for the turret-gul pieces, you yourself admit that they are nowadays always attributed to the Tekke because they are always open right; had any been made by the Salor there would presumably have been at least some that are open left.
While we are at it, I wish to question the ཐ% os Salors open to the right assertion. I don't know where it comes from and what evidence there is for it. I have grave doubts about it accuracy. But this is a side issue that has no bearing on the main point.

Regards, Yon

The fact that we attribute certain things to the Tekke is not a matter of dispute. The question is, on what basis did we arrive at the position of doing so on structural grounds? The case of the 6 turreted gul juvals being Tekke is probably pretty solid, since, as far as I know, every last one of them that anyone has looked at is asymmetric right knotted and most of the 3 gul juvals are asymmetric left knotted and given Salor attributions. The 3 gul asymmetric right juvals are often given Tekke attribution - I think this may often be an error.

Jon Thompson's Turkmen describes Salor stuff as usually asymmetric left, but with exceptions. Jourdan's book says essentially the same thing. I can't put my hands or brain on the source of the 20% figure for the asymmetric right Salor, but I believe it was something George O'Bannon wrote. But let's just ignore the actual percentage. Thompson and Jourdan both say that there are exceptions to the Salor = Open Left equation. How could they conceivably have arrived at that conclusion except by using design, etc. as a basis for attribution? Admittedly, it could be a statement made by one person and repeated through the literature.

It would b e good to know on what foundation structure based attribution rests if it isn't the design, palette, motif triad.

The eagle gul group has itself been subdivided into three groups on the basis of structure, and the original identification was based on the fact that the structure of eagle gul carpets differed from most Yomud stuff. There are rather few specimens, and one of O'Bannon's criticisms of the whole business at one point was that it seemed as though a new "group" was being created every time a structural variant was discovered.

I think examining the rocks on which our castles are built is a worthwhile exercise, bythe way. Setting up straw men and seeing whether we can knock them down is the fundamental paradigm of scientific inquiry.

Regards,

Steve Price

Steve, if you are making any point at all, it is (as far as I can tell) that our structural information is derived from analyzing rugs whose original attribution comes from their looks. That is true to a large extent, but as my examples show it is not quite so simple a process. My examples show that when designs are common to several sources (Tekke and Salor, Shirvan and Daghestan) then classification relies on structural clues that are derived from other weavings with designs that are thought to be unique to the various sources. But, as the Eagle-gul pieces show, new classes can be surmised purely from structural analysis, although attributing them may be difficult or impossible (Azadi tries valiantly with the Eagle-guls). And, btw, new Eagle-gul pieces are surfacing continuosly; I think there are many more of them than originally suspected.

Regards, Yon

Let me respond only to one part of your posting :

Until some time ago, all prayer rugs of a certain familiar design were called 'Daghestan.' Later it was concluded that those with a flat back were Shirvans, with slightly depressed warps were Kubas, and only the ones with strongly depressed warps were actual Daghestans. I don't know where the latest scholarship stands on this issue, but I am sure that structure is still the most reliable tool for classifying these weavings.

I don't know what the most recent thinking is on this, either, but I'd ask the following question: How did anyone arrive at the conclusion that those structural characteristics were determinative for the geographic attributions? In other words, what makes you (and nearly everyone else, including me sometimes) so confident that the old attributions were wrong and that the new ones are right? I don't think that's an easy question to answer, and I don't think the question is trivial. It really gets to the heart of the matter that I'm raising.

Regards,

Steve Price

"My examples show that when designs are common to several sources (Tekke and Salor, Shirvan and Daghestan) then classification relies on structural clues that are derived from other weavings with designs that are thought to be unique to the various sources." In other words, if we observe that all authenticated Shirvans have flat backs and all authenticated Daghestan have ridged back, then we may have a handle on classifying the 'Daghestan' design rugs.

The next question is how is the origin of a rug authenticated? Surely not purely, or even primarily, on the basis of marketplace lore. Relying on the marketplace leaves us with Cabistans and their like. Until quite recently, a prominent Boston-area dealer was exhibiting a Sevan Kazaks in local museums with attribution to "... Kazak woven in Central Asia ... the product of the Kirghiz tribes ..."

In the end, attribution must be based on field work, trustworthy archival material (records of original purchases, e.g.), and careful scholarly analysis of all available evidence. Perhaps a Salon on the subject of 'how are robust attributions made' would be useful.

Regards, Yon

I agree with essentially every word in your last post. Most attributions can be traced to little more than marketplace folklore of which designs, palettes and motifs originated in which areas or tribes, and we both recognize that this is a pretty shaky foundation for any edifice.

I think the matter of what it takes to make a robust attribution, and how that relates to what we normally do on a day to day basis, is extremely important to ruggies, and almost never discussed. In my own obtuse way, I intended for this Salon to more or less lean in that direction.

If you would agree to host a Salon facing the matter more head-on(and assuming that you haven't totally lost patience with me), I think it would be a productive undertaking.

Regards,

Steve Price

Regards, Yon

Once we have a group of rugs identified as having particular features, how to we connect them to particular weavers or to a particular place?

Yon appears to commit to hosting a future salon on this question.

Meanwhile, it might be good to point to one methodology used recently on Caucasian rugs in particular.

In their "Caucasian Carpets and Covers" Wright and Wertime introduce the notion of an "anchor piece" or "anchor rug" for use in establishing provenance.

They say: "Establishing the provenance (ed. of an "anchor piece") requires the testimony of someone who had first-hand knowledge of its origin. The written and photographic records of pre-revolutionary Russia provide this witness. The photographs can be matched with extant textiles. While woven structure can rarely be determined from a photograph, a close correspondence in design between a textile in a photograph and one in hand permits the linking of structure to textile. Such identities then can be confirmed by the absence of objects similar in appearance, but with different structures."

Now one can tussle, too, with this methodology (it is has always seemed weaker to me at its inferential end) but it is a clear method for attempting to establish provenance more systematically.

Steve, will have noticed that design is central to it and in fact prior.

Regards,

R. John Howe

Regards, Yon

Interesting that John brings up Wright and Wertime's Caucasian Carpets and Covers and notes that the foundation it uses for attribution is design, palette and motifs. More recently, John Wertime published Soumak Bags, an attempt to make attributions of flatwoven bags on the basis of their similarities (in design, palette and motif) to rugs and carpets from specific areas in NW Persia and the Caucasus.

Both books are generally regarded as excellent, scholarly expositions with major advances in attribution methodology. I happen to agree. I think they are the best we've got on the subjects.

But they do illustrate nicely the thesis of this Salon: the attribution criteria on which we rely usually rest on a foundation of those characteristics that can be seen in a typical photo in a Sotheby's catalog, and attributions from structure, in fact, derive from them in almost all cases. And, most important, that doing so includes the work of respectable scholars working today, not only on marketplace folklore from 50 to 100 years ago (although much attribution is from that).

The Daghestan/Kuba/Shirvan attributions of the white ground prayer rugs with flowers in lattices are still more a matter of faith than of hard evidence. As Yon pointed out, the issue of authentication of these pieces has really not been properly addressed.

Regards,

Steve Price

You wrote in part above:

"John, the Daghestan example shows how unreliable this methodology can be. It simply cannot cope with cases of identical design coming from different sources and sporting different structures."

My thought:

I'm caught without my books to cover the likely nakedness of my unassisted thought, but I think Wright and Wertime would reply that they don't ask this methodology to do more than it can.

It is for this reason, I think, that they advise that "Talish" pieces are more accurately described as of the "Talish style." The varying structures of some of them apparently suggest that "Talish" designs were woven in a number of different locations.

Someone with access to the book should look to see what moves they make on Yon's Dagestan challenge.

Regards,

R. John Howe

Regards, Yon

Your statement, ...when designs are unique to a source they serve as a good identifier and when they are not they don't is true, but not very useful. The problem, of course, is that knowing whether a design is unique to a source is impossible, or nearly so.

Regards,

Steve Price

I think I'd state the Wright-Wertime thesis in this area differently.

If a Caucasian rug in hand has a design that closely resembles that in of a rug photographed in a know Caucasian location before 1917 (the photographer saying that he knew that it was made there) AND the known examples of rugs with that design have common structural characteristics (that is, they do not differ much from one another in this respect) then we MAY be in a position to infer (and an inference is certainly is) that the rug in the picture has the same structure as the rug in hand and, more to the point, that the rug in hand was made where the rug in the photo was.

I think the inference Wright and Wertime recommend is shaky enough as they state it and doesn't require parody.

Am I the only one with a copy of "Caucasian Carpets and Covers?" No wonder it's been remaindered for so long.

Regards,

R. John Howe

I took a look at the Wright-Wertime treatment of the Daghestan area last night and came away thinking that it is richer and more complex and considered than Yon's one-line rejection of my description of their methodology merits.

I won't summarize what WW say but do recommend that interested folks read it.

On Yon's side of things, I should acknowledge that another Caucasian authority, who has been watching this thread, wrote me on the side last night to say that he is not much impressed with this book either. He was fairly harsh about it.

So read and decide for yourselves.

Regards,

R. John Howe

Regards, Yon

First, to John: I have copy of Wright/Wertime, but not in my office and didn't have time to look up what it had to say about Daghestan prayer rugs last night. For the benefit of people like me and for those unfortunates who don't have the book at all, would you be good enough to post a brief summary of the content that you are suggesting that we all read?

Second, to John: I'm aware of criticisms, sometimes harsh, of the Wright/Wertime book. While I think there are flaws, I also think the standard of comparison has to be what else is out there. It fares very well by that standard; it is beautifully produced and illustrated, has an well-documented ethnographic history of the area, and presents an imaginative and original approach to attribution from photos of prototypes.

Third, to Yon: I understand that you intend the statement, ...when designs are unique to a source they serve as a good identifier and when they are not they don't to be simply a terse summary of the state of attribution by design, etc. It must be just as true, then, to state that when a structure is unique to a source it serves as a good identifier and when it is not, it doesn't. Where does that leave us?

Regards,

Steve Pride

Regards, Yon

Editor Note: This thread continues on another page. The button leading to it is at the lower right side of your monitor.

That's true only if we can authenticate the criteria. As it turns out, the authentication of the structural criteria is usually (although not always) by comparison of pieces with similar design, palette and motif. The authentication of those criteria often cannot be traced much beyond a faith in marketplace wisdom.

Regards,

Steve Price

What I would say at the point in the sequence of indicators that you list above is, not that we have identified a "Salor," but that we have identified "a group of rugs that meet a particular set of indicators."

It always requires a separate demonstration to show who the group now identified is woven by a particular group or at a particular location. (This is likely Steve's point too in the post immediately above this one. I'm just stating it in my own words for confirmation.)

This is the point that Murray Eiland has argued with Pinner for years, resisting (he seems no longer to be protesting out loud in his most recent edition of his comprehensive guide) the notion spread by Thompson and Pinner that we have demonstrated adequately that the "Salors" are the group that wove the "S" group pieces.

So while I support Steve's general salon thesis only at the level of a dichotomy of indicators seen at about two feet and those that require a closer look, (I still think structural factors should in general be weighted more heavily because they change more slowly) I don't think we usually have demonstrated the links between a given group of rugs (identified by reference to a number of indicators) and the either a particular set of weavers or the place(s) where they were woven.

Regards,

R. John Howe

Regards, Yon

I'm afraid I've not made my position clear at all. Identification of the actual source of a group of rugs - authentication - is not the same problem as identifying the existence of a group.

The Dokhtor-i-Qazi Belouch is an example. Those are rugs easily recognizable from their greatly similar designs (layouts), palettes and motifs. They are believed to have a common origin. That origin, in specific terms, is unknown. That is, beyond assigning them to Belouch group weavings, we really don't know who made them or where they were made.

It turns out that they have similar structures, too, but not so unique as to define the group on structure alone.

Now, suppose that tomorrow morning someone send us a photo of what is obviously a Dokhtor-i-Qazi prayer rug. That is, a rug with the border, field motifs, mihrab shape, minor borders, checkerboard selvage, palette (including the very unusual cobalt blue) and sequence of horizontal stripes in the kilim ends that was the same as all the members of the group. BUT, this rug, unlike the others, has symmetric knots. Would we call it Dokhtor-i-Qazi? I think we would; maybe not. Suppose we had 100 specimens, 75 with asymmetric knots and 25 with symmetric knots. Would we consider these to be two different groups or all one group? I am pretty sure we'd call it all one group, that we'd base our attribution decision on the design, not the structure. Why do I think that? It's because that's exactly what we do now with Yomud, Ersari, and other central Asian groups. Am I sure we'd be correct in doing so? Not at all.

Regards,

Steve Price

---

Yon Bard (08-26-2001 08:58 p.m.):
Steve, while the 'eagle gul' pieces are named after the eagle gul, there are quite a few that are assigned to these groups purely on the basis of structure, with no trace of eagle guls. This applies to all the non-main carpet pieces, and some main carpets as well.

As for the turret-gul pieces, you yourself admit that they are nowadays always attributed to the Tekke because they are always open right; had any been made by the Salor there would presumably have been at least some that are open left.
While we are at it, I wish to question the ཐ% os Salors open to the right assertion. I don't know where it comes from and what evidence there is for it. I have grave doubts about it accuracy. But this is a side issue that has no bearing on the main point.

Regards, Yon

---

Let me introduce the idea that changes in weave pattern within a single tribe like the Eagle Gull weavings might actually represent three or four hundred years of weaving history compressed into a few dozens of pieces. As far as the 20% of all Salor weavings being open right, further analysis of these pieces has revealed that a large portion of them are Tekke, very very old, with Salor iconography. Taking a close look at these pieces I have concluded that there is a 16th century Tekke aesthetic that is patterned off of but not identical to the Arabatchi(main carpet main gull) and the proto Eagle gull tribe( minor element and chuval gull drawing of Hoffmeisters 16th century tekke torba, whoever they were. The evidence shows that in the 17th century the few Tekke weavings we have from this period are 100% Salor in iconography. This leads me to speculate that the Salor were very preeminant in the 17th century and times were rough. Early 18th century Tekke weavings have very complex representations in visual depth and are extremely finely woven with glorious colors. These are from the good days at Khiva when clear waters flowed and the sheep were fat and excellent. In conclusion I speculate that the Salor were always weaving asymmetric open left and always with at least some warp depression, often deeply depressed. Going further out on this limb I imagine the Eagle Gull were a large and powerful tribe in the middle ages of our history and that some of the clans did integrate into the Yomud and this in the latter 18th century onward into the 19th. century. Their earliest weavings, Type one mains, are very very old and have amazingly complex constructions pointing to a very long incubation. The rigid standards of these earliest carpets are relaxed as they become more recent. In the 19th century we seem to be seeing a Yomud aligned eagle gull group and a free and independent eagle gull clan, possibly the last one. Neither group has passed a large number of weavings along to us. Jim Allen

Regards, Yon

I'm not sure which of my assertions you believe are refuted by the the evolution of attributions of flower-in-lattice prayer rugs. Once upon a time they were all thought to be made in Daghestan, since there is a tradition of making flower-in-lattice rugs in Daghestan. Subsequently, thinking on this changed and we now believe we can tell Daghestan from Shirvan from Kuba flower-in-lattice rugs, mostly on structural grounds (but not entirely - one of the major clues to distinguishing among the three classes is border design). The evidence that we now have it right is almost nonexistent, and for that reason it doesn't make a compelling argument for or against any assertion by me or by anyone else.

But, just for ha-ha, let's assume that the current view really is right - that we really can tell Daghestan from Kuba from Shirvan flower-in-lattice prayer rugs, and that it can be done entirely on structural grounds. In what way would this prove that most structure-based attribution rests on design-based attribution, which is the central point to my Salon essay? Even if I grant that it is one of the examples that is not part of "most", this still leaves the majority of the cases as within the rule. Wright/Wertime's Caucasian Carpets and Covers and Wertime's Soumak Bags, both important contemporary works dealing with attribution of large numbers of groups of textiles, rest their methods almost entirely on design, palette and motif. Not everyone thinks these are great works, and I have seen criticism of various aspects of them. But the closest thing I've seen to a criticism of the use of what we might call "macro" visible properties for attributions is that there have been objections raised to the so-called anchor pieces in the Wright and Wertime book because they are very late - mostly 20th century. I haven't seen even a hint to suggest that anyone objected to the attribution methodology on other grounds. That doesn't mean that nobody has any objections, only that if anyone has articulated them, I'm not aware of it.

And if most respectable authorities have no objection to the use of these properties as primary attribution criteria, my assertion that this is the custom is not refuted by the existence of a few exceptions (if, indeed, those really are exceptions).

Regards,

Steve Price

Regards, Yon

That is close, but there's just a little more to the whole point. Let me briefly complete it here so that I don't waste too much more of your time or mine. So far we're together on this much: ...design is used for attribution most, but not all of the time, and that much, but not all, of the association between structure and origin is based on design.

If that's true, and as nearly as I can tell we (you and I) agree that it is, then there's a corollary that springs from it. The corollary is that much, but not all, of the attribution based on associations between structure and origin cannot be more reliable than the attribution based on the design all by itself. The conclusion I reached from this simple truth in the Salon essay is that the frequent assertion that the most reliable basis for attribution is structure, not design (plus palette and motifs) cannot be correct.

I don't think this is is something everyone knew all along.

Regards,

Steve Price

Regards, Yon

...one pair of rugs identical in design but different in structure, and another pair identical in structure but different in design. Which pair is more likely to come from the same weaving group? I think most experts nowadays would agree that it's the second pair. If that is correct, then structure IS the more reliable criterion.

It is my impression that this is, indeed, the position most experts would take. In fact, if the conventional wisdom before last week held that my thesis is correct, there would have been little point in presenting it.

I don't see how the prevalence of expert opinion proves which of the alternatives in your hypothetical case is correct without assuming that expert opinion is the criterion by which to test truth.

I think expert opinion is usually trustworthy, of course, but when the epistemology underlying expert opinion is flawed (as it appears to be in the topic under discussion), expert opinion is not reliable. Expert opinion for centuries was that the earth was flat. The epistemological basis for that opinion was flawed, and the opinion was incorrect. At least, that's my (inexpert) opinion.

Regards,

Steve Price

Regards, Yon

You are correct, the proposition is either true or false. That is not a matter on which we disagree. The question is, how can we know whether it is true or false?

I've tried to offer an answer, along with an explanation of how I got to it. You think my answer is incorrrect, but the only explanation I've seen is that expert opinion is on your side. You appear to agree that this is not very compelling evidence.

If you believe that the answer lies in painstaking scientific investigation, I would ask, painstaking scientific investigation of what? And if you don't think reasoning from what that investigation provides is the route to an answer, what do you intend to do with that data?

I'm not trying to bait or annoy you, although obviously that's what's happening. Maybe the whole discussion is a waste of time, but I suspect that there are people out there who are learning something from it. If we've done nothing more than make some of them examine the basis for what they believe to be true of the process of attribution, I think that we've done something useful.

You are a forceful and thoughtful advocate of your point of view, and I suspect that many readers appreciate that as much as I do. And for those who don't, I offer the following entertainment:

Regards,

Steve Price

Regards, Yon

I am going out the door on a week's vacation and I hesitate to intrude on the "yes it is," no, it isn't sequence above, but I want to suggest to Steve that he is pointedly ignoring why many folks now feel, not that structure by itself is a better indicator of the presence of a distinctive group of weavings, but that structure is generally a more reliable indicator comparatively than is design in particular.

This reason is that the potential costs to a traditional rug weaver of changing traditional structural features of weavings are seen to be much higher than those associated with weaving a new design.

So this is not just a case of more people believing one thing than those who believe another (although you must be the moderator in this salon, fomenting a little controversy, because I've heard you argue precisely the reverse position on such a question, for example in your critiques of some of Elena Tzareva's attributions). It is that they do so BECAUSE it appears that structural features change more slowly than designs do.

Now it may be that your real thesis is that all the factors that can be seen from two feet away, combine to be more predictive than structure or even than when we examine structure our judgments are already contaminated with information we unavoidably have about more visible indicators but that seems somewhat different than the narrower thesis that I seem sometimes to hear you argue in these exchanges.

Regards,

R. John Howe

First, let me say that if I don't think the same things or the same way as I did at some time in the past, that simply means that either I was wrong then but and right now, that I'm wrong now but was right then, or that I was wrong then and am wrong again. It's not against the rules to change what you believe when faced with new information.

Second, I think it might be worthwhile to quote what I think of as the essence of the Salon thesis, since this thread is now longer than that original essay and has gone off in several directions that tend, to some extent, to distract us from the main point (some might say that's for the best).

Here are the two most important statements,

First, the basis for the belief that structure trumps design, etc.:
1. "...someone who has been weaving rugs since childhood is much more likely to adopt new layouts, motifs and palettes than to change the mechanics of how she weaves."

Next, the heart of my argument,
2. "I suggest that layout, motifs and palette may be better attribution criteria than structure. I will argue that the foundation of structure based attribution is layout, motif and palette based attribution and no method can be more reliable than the foundation on which it rests.

Please, if you will, notice that there is nothing in the second statement that denies the importance or preeminence of scholarly research in documents or in field work, when such exists. Most of the time it doesn't, unfortunately. And even when it does, many times it simply associates the design, palette and motifs with a geographic area.

I do not hold an absolutist view that design, etc. is the always more reliable than structure, but I do reject the absolutist view that that structure is always more reliable than design, etc. The second italicized paragraph is my reason.

I don't think repeating my argument makes it more convincing, and I did so here just as a reminder of what it actually is.

Regards,

Steve Price

Thanks for restating the hypothesis. I believe that you are saying that the study of the structure of rugs, in order to determine their attribution, is only necessary because (and follows from the fact that) sometimes the structure of a rug is not the same as all of the others visually like it. The historical attributions based on marketplace lore do not explain some of the structural differences, therefore a structural comparison may allow us to "differentiate" rugs that appear otherwise visually similar.
It does not necessarily lead us to a more likely actual attribution, but just a confirmation that there are structural differences within a group of weavings that are visually similar and that have historically been "lumped together" because of their visual similarities.
There is no logical way to argue otherwise, therefore there is no refuting the logic of your argument that "The epistemology (or method) of attribution based on structure" is directly derived from the earlier visual method of attribution.
This does not negate the need for structural descriptions which allow this differentiation. The differentiation and sub-categorization of rugs within a larger set of weavings allows the field of rug studies to progress.
I am certain that you do not argue for neglecting or discarding the process of structural analysis of rugs. You are just saying that the structural analysis follows from the visual analysis.

Patrick Weiler

Thanks for making my muddy explanations more clear. You've got it the way I intended it to be, although I would add that the design, etc. being the base for the inferences on structure is only true most of the time, there are some exceptions.

I'm going to take another stab at unmuddying the water a little in a new thread.

Steve Price