Posted by Daniel Deschuyteneer on 06-01-2003 02:41 PM:


At the end of the 19th century, Karabagh, the most southern part of central Transcaucasia, was inhabited by several groups of people, the most important of them being the Armenians, the Azeri and the Kurds. Except for items containing Armenian inscriptions or dates, it is very difficult to attribute with some certainty, the several utilitarian we come across to a specific group.

Some weeks ago a restorer friend phoned me, telling me, he had to restore a near complete Caucasian mafrash (bedding bag) woven “without wefts” between the soumak rows which should get my interest.

I visited his shop some days later, having in mind that it shouldn’t be a Caucasian mafrash, but well, a weftless soumak Eastern Anatolian or NW Persian Kurdish mafrash, as weftless soumak is a Kurdish practice. We have already discussed this extremely antique technique used by Kurdish weavers in earlier Salons presented by Michael Wendorf or by me.

When I saw it for the first time, three panels where still attached together: the front panel (seen here) and the to side panels. Only remnants of the back panel were still present and I received these fragments to examine them more closely.

The colors were very difficult to capture and are not perfectly accurate. The reds especially are too much saturate. Get a look to the direct scan to get a better idea of the colors.

I directly recognize it was a Karabagh piece, because a near exactly similar piece is illustrated in the John T. Wertime book: Sumak bags of NW Persia and Transcaucasia – plate 110 – page 180 - According to Wertime many Qarabagh soumak bedding bags came into the West in the hands of Armenian immigrants who were prolific weavers in this area, but to my knowledge they didn’t use the weftless soumak technique.

So “IF” this piece would have been woven with weftless soumak it would have been woven by another ethnic group.

I was exited and I began to examine it. Unfortunately the mafrash was so finely woven that I couldn’t examine it correctly with my usual lenses. I was not sure, but its structure looked nevertheless unusual.

Studying the remnants of the back panels with magnifying lenses I finally discovered it had a very unusual structure, situated “midway” between usual soumak weaving and weftless soumak.

Here is a direct scan from the front having been magnified 15 times .

Rows of sturdy plain 2/1 soumak wrapping are well visible and red wefts visible along the sides are inserted after lots of rows of wrapping.

Looking now at a 15 times magnified scan of the back things become clearer.

In fact, in this area, red wefts are inserted after 2, 3 or 4 rows of wrapping.

Provocatively and to imitate Alberto LEVI who used for the first time the label “proto Kurdish” I have labeled this structure “proto-soumak” as it suggests (it’s only a guess) that such mafrash may well have been woven by Kurdish hands, knowing the weftless technique, instead of Armenians or, by weavers who learned it from Kurds.

Here is another area showing in the upper part (four first rows of wrapping) usual 2/1 soumak with red wefts inserted after each row of wrapping and than in the lower part a large “pure” weftless soumak area.

It would be interesting to know if readers do have related or similar pieces showing the same unusual structure. Is it a characteristic of this group?

Also I am interested to know why the weaver used this “hybrid technique”.

As I said, this mafrash is very tightly woven, about 40 extra weft wrapping and 10-12 wefts per inch.

I am thinking that the weaver didn’t insert a weft after each row of soumak as usually done, to give this bag a supple handling, as it is, but there are may be other reasons I am not knowing. This piece is so much tightly woven that the rows of “weftless soumak” were certainly not woven to spare time as it has been suggested to explain why Kurds used weftless soumak.

Technical analysis:

Sizes : 105cm x 47 cm (originally it was about 110 x 50 cm)

Yarn spin: Z

Technique: 2/1 plain hybrid “proto – soumak”

Wrapping yarns: 2 ply wool and 3 ply hand spun undyed cotton, 40/pi

Warps: 2 ply light brown wool, 16pi

Wefts: red wool singles, one shoot after 2 to 8 rows of soumak wrapping, about 10-12 pi

Marla tells us (Woven structure page 65) that in some soumak weavings we may find short extra rows of wrapping before one weft is inserted, where the weaver has tried to straighten her work and that this occurs most often when both wool and (thicker) cotton yarns are used for wrapping, since these yarns are likely to vary in size and compact differently.

In this piece the “hybrid” technique has been used for other purposes and I have larger pictures that I can send privately to anyone who would be interested.

I would be glad to share your thoughts and opinions.



Posted by Sue Zimerman on 06-04-2003 11:37 AM:

Hi Daniel,

The lofty cotton, over time and with abrasion caused by usage, would cause structural weakening of the bag. In addition to compensating for yarn diameters, compartmentalizing the weaving into horizontal areas for ease in controlling the structure/design, the wefts, being wool, would lend stability, as the abrasion occurred, by interlocking microscopically with the warps.

Another purpose served by adding wefts rather than, or in conjunction with, additional rows of compensating soumak, is that the structure/design could be controlled better by beating on a continuous weft than on discontinuous soumak design elements. Too, if the wefts were to be damaged by heavy beating it would cause less of a problem, in the long run, than beating directly on the soumak itself.

I imagine that there are more wefts added in the areas where the most cotton was used. Is this so? Sue

Posted by Deschuyteneer Daniel on 06-04-2003 03:15 PM:

Hi Sue,
Thanks for your interest. The wefts seem to be continuous all the way across the piece It seems there isn't any planned variation of the number of wefts and I don't see any variations of their numbers according to changes in the design or colors.

Posted by Michael_Wendorf on 06-04-2003 05:12 PM:

simple or weftless and compound weft-wrapping

Dear Daniel:

Thanks to you and your restorer friend for your careful deconstruction of this interesting mafrash. I immediately recognize the mafrash as I have seen several near identical pieces over the past years. The distinctive design and coloration is consistent in these pieces. In addition to the example you reference in Wertime's Sumak Bags (plate 110) there is another illustrated in Tanavoli's Shahsavan as plate 57. The piece in Tanavoli's Shahsavan is two front panels sewn together with the basic design called "crab shaped stars." The structure is:

warp: wool Z3S, dyed , light brown with some salt and pepper areas (brown and white);
weft: ground - wool Z2S, undyed, light brown. Wrapping weft -wool Z2S dyed and with some areas of cotton that is Z3S.
Structure: predominantly extra weft wrapping patterning a plain weave ground with one ground weft after each wrapping weft, horizontal 2/1, diagonal 2/1, 3/2, vertical 1/1, 2/2. The geographic attribution is Moghan.

The piece in Wertime's Sumak Bags has no structural details.

I find your description of the fine and tight weave a great help in understanding this mafrash. In fact, your description explains what you are observing. In the Appendice to Wertime's Sumak Bags he briefly discusses examples where a ground weft is lacking in places. He does this by referencing the famous cruciform soumak bag owned by Wendel Swan. Wertime writes: "Sometimes the weaver dispenses with the ground wefts altogether and employs only the wrapping wefts for several or more rows, which thus become very tightly compacted (plate 74). Such a structure is known as simple weft-wrapping. It is used particularly in narrow horizontal borders or solid lines going edge to edge." See p. 224 of Sumak Bags.

Of course, weft-wrapping in northwestern Iran and Transcaucasia is overwhelmingly a compound weave. By compound weave I mean that in addition to the warps it uses a weft that may be understood or called variously a ground, structural, foundation or interlacing weft. It also uses a supplementary or extra weft to wrap or pattern the warps. What Wertime calls simple weft wrapping and what Marla Mallett, George O'Bannon and I have called weftless soumak is different. As a simple or weftless form of weft-wrapping it uses no ground, structural, foundation or interlacing weft. Weftless soumak is the most ancient soumak construction. This weftless soumak structure is primarily known for its use by early neolithic weavers weaving with bast fibers and by Kurdish weavers in eastern Anatolia much later. It has a distinctive look and feel that is different from this mafrash panel.

I think that your use of the term hybrid is a good way to think of this mafrash. It is not Kurdish in structure, coloration etc. I think the Qarabagh attribution geographically very likely. I also think that we cannot think of this as weftless soumak insofar as that structure is used to describe weavings done completely in that structure. We go back to the tight weave, the weaver wanted a weaving that was very tightly compacted, this was achieved by omitting the ground wefts in places. It is a simple, but elegant explanation.

As for the provovative use of a term like proto-soumak, this is only to provoke discussion I think insofar as the term has no meaning in connection with this weaving.

Thank you for the provocation, Michael

Posted by Marla Mallett on 06-05-2003 01:08 PM:

Daniel has said that he has sent me some fragments of his mafrash to analyze, but I’ve not yet received them. I can, however, make a couple of remarks about the soumak techniques that are being discussed.

The major critical difference between Kurdish weftless soumak pieces and a piece such as this mafrash, is that the wrapping has been worked in completely different ways. On Kurdish pieces, the artisan typically has wrapped the warps in one direction, then has floated her wrapping yarn backwards over the just completed section to wrap the next short row in the same direction. (WOVEN STRUCTURES, Figs. 5.19 and 5.20, p. 68). To do this means that all design segments must be relatively narrow. It’s these backward floats that tend to pull inward and so give each section a distinctive “rounded” or “padded” sort of look on the front. The important thing is that as with slit tapestry, the weaver works on individual design sections separately, completing each little area before moving on to do another section of the piece. She does NOT weave the piece in consecutive rows all across the loom. Thus there’s no way for intermittent wefts to be used with the Kurdish wrapping technique. The Kurdish designs used for soumak are quite limited and distinctive; they are very closely related to slit tapestry motifs. I have one detail photo of a Kurdish bag in WOVEN STRUCTURES (Fig. 5.31, page 71) that shows an area in which the weaver has experimentally combined soumak wrapping and tapestry. Often Kurdish weftless soumak design layouts are banded, as plain-weave bands throughout the pieces offer the stability needed in the fabric.

The soumak wrapping technique displayed in the ancient Neolithic Anatolian wrapped net fabric found at Catal Huyuk is much closer to that in Daniel’s Caucasian mafrash and to typical Shahsevan pieces, than to the Kurdish East Anatolian weftless soumak pieces. The Neolithic fragment shows conventional wrapping of sequential rows across the fabric, with reversals at the selvage.

Without examining Daniel’s mafrash fragments, I’m assuming that the omission of part of the “ground wefts” was simply to allow for a more compact weave. The short rows of wrapping reversed, with the weaver working in both directions, changing her process in alternate rows to produce an un-countered surface. She certainly used enough ground wefts to produce a stable fabric and close the vertical slits that otherwise would have resulted from long verticals in the design. But that’s not the critical difference between the two approaches to soumak wrapping. The important factor is that in the Kurdish pieces, the weaver used a slit-tapestry approach, weaving all of the design parts separately, and composing designs specifically for this process. It’s an important “technical” difference, as well as a “structural” one.


Posted by Michael_Wendorf on 06-05-2003 02:35 PM:

a more compacted weave

Dear Daniel and Marla:

Yes, the desire to create a compacted weave explains the ommission of the "ground wefts" in some areas in the mafrash panel and other Caucasian soumaks. And Marla's explanation of the difference between roughly contemporaneous Kurdish weftless soumak weavings and Caucasian soumak weavings such as the mafrash panel expands nicely from my conclusory remarks, thank you.

Marla also writes that the soumak wrapping technique displayed in the ancient Neolithic Anatolian wrapped "net-like fabrics" (to quote Burnham) found at Catal Huyuk is much closer to that in Daniel's Caucasian mafrash than to the Kurdish East Anatolian weftless soumak pieces. I find this conclusion quite surprising. For one thing, the ancient Catal Huyuk "net-like" fragments we are now discussing and comparing are fragmentary and heavily carbonized. They were woven without any pattern, at least no discernable pattern. As a result, there are no color changes or design segments that would allow for any apple to apple technical comparison and it is fairly difficult to imply or infer that the wrapping on weftless soumak pieces that are patterned is or would be different from weftless soumak pieces that are unpatterned. Moreover, they also appear to have been woven entirely in weftless soumak, - - - not partly and not by omission. The ancient Neolithic Anatolian Catal Huyuk fragments that do exist are woven entirely as a simple wrapped structure (no ground wefts at all) with threads plied Z2S or with Z singles with one transverse thread (the wrapping weft) going over two vertical threads or warps and back under one.

Perhaps this simple wrapping of an unpatterned fabric is what Marla refers to as "conventional wrapping of sequential rows across the fabric with reversals at the selvedge." But again what would one expect in an unpatterned fabric? I fail to see how this fact makes the technique and structure closer to the mafrash panel type of weaving where some ground wefts are omitted to make the the weaving itself more compact.

As for the patterned east Anatolian Kurdish weftless soumaks, it is true that they are not woven in consecutive rows across the loom and that there is no way for what Marla here refers to as intermittent ground wefts. As a result, just as with the early Neolithic Anatolian fragments, ground wefts are not absent by omission or to create a more compact weave --- they do not exist at all. Additionally while some examples are banded with plain weave as Marla states, there are many examples that are not. Finally, while the designs/motifs of the Kurdish weftless soumak pieces are related to slit-tapestry it may be that it is only because the two techniques are worked so similarly and are both restrictive in nature when a weaver wishes to create a pattern.

I think any comparison between the ancient Neolithic fragments and a Caucasian mafrash with grounds omitted in some areas to create a more compact weave a stretch. Perhaps it is also a stretch to compare the Kurdish east Anatolian weftless soumaks. In any event, I think the Kurdish pieces, which are woven as weftless soumak weavings rather than as compound weft wrapping with some ground wefts omitted, deserve to be understood and discussed on their own terms and not in comparison to weavings where the structure is a convenience or a tool.

But I guess this opinion is well known. Now i am going for a drive in my 512BB.

Cheers, Michael

Posted by Sue Zimmerman on 06-05-2003 03:10 PM:

Hi Daniel,

I can't imagine why anyone would not be interested. Thank you for putting this weaving out there. Your photos and stats make it an excellent piece to learn from. I made my posted assessment based on the large "pure" weftless areas in your photo #3. I think it is one of those horizontal boarders or solid lines going from edge to edge that Michael sites Wertime as speaking of. I think these design/structural features would only be necessary, from an engineering standpoint, when a soumak weaving included a lot of fragile cotton. If you have enough of a fragment to include another area like the one in your photo, ( it will be one with little or no cotton), if Wertime is right, as it is reasonable to assume he is, it will be weftless too.

Hi Michael,

Thank you for the additional good info. I agree with you that the weaver's intent was to weave a tightly woven mafrash, and that this was accomplished in a simple and elegant way. I think though, too, that this hybrid technique would not have been necessary had it not been woven with a mixture of wool and cotton yarn, for reasons I stated in my last post. I think we can agree that the weaving was built to last. I think of the weftless bands as a design/structural feature used by the weaver to solve a problem created by the nature of the cotton used, much as we, today, use strapping tape on cardboard containers destined for the post office. If I am right, the weaver's solution was even more simple and elegant then it first appears!

Another thing I noticed. In Photo #2, under the cross' horizontal extension, if you follow the lightest row of blue soumak to the right, I think you can see it slants and that a "wedge" of extra rows of blue soumak has been added to straighten the work. Sue

Posted by Marla Mallett on 06-05-2003 04:24 PM:

Dear Michael,

In the development of new fiber structures, it is unwise to forget the overriding importance of the PROCESS, the TECHNIQUE, and only look at the resulting STRUCTURE. It is the particular wrapping PROCESS that I was attempting to stress. Perhaps it is necessary to have spent a significant amount of time actually using these textile-making processes to realize the importance of the process itself on designing, and upon the development of any fabric or fabric tradition.

The Catal Huyuk wrapped textile was a NET, with widely spaced rows of wrapping, and of course no pattern. But the process involved wrapping sequential rows, and reversing the direction at each selvage for alternating rows. If Harold Burnham’s diagram is correct, the wrapping was done without putting the wrapping yarn through a self-made loop while wrapping each warp, and this same movement was followed in every row, thus making “countered” rows. There is so little parallel between this and the process used by the Kurds with their “sectional” wrapping in slit-tapestry fashion with continual backward floats so as to always work in the same direction that comparisons between the Kurdish structure and the Neolithic one actually make little sense to me. The similarities are superficial. And yes, in this respect I do find more direct parallels with the way that typical Shahsevan soumak is actually PRODUCED (whether or not any wefts are omitted) and the Neolithic net-making—if any comparisons at all are indeed appropriate. Of course those Neolithic NETS had no plain-weave wefts. Have you ever seen any kind of NET with plain-weave interlacing? Most impractical.


Posted by Sue Zimmerman on 06-05-2003 06:02 PM:

In the development of new structures of ANY kind, it is wise to remember the overriding importance of the limitations and properties of materials to be used. Sue

Posted by R. John Howe on 06-05-2003 06:49 PM:

Sue -

I don't think that this is a debate.

I doubt that Marla would under-rate the character of the materials used in weaving, since she is herself a weaver.

But one of the chief recommendations she has made in most of her writing is that it is necessary when we examine structural aspects of weavings, not just to describe what we see, but to take on the perspective of the weaver and to grapple with what she as called above, the "process" of the weaving. Surprises are likely whenever we assume this perspective.

I don't think, for example, that Michael Wendorf would debate at all the distinction Marla has made between the character of the Catal Huyuk weavings being referred to here and the weftless sumak of the Kurds, if he grappled, even in a beginning way, with what the weavers did when they made these alternative structures.

This is the sort of thing that Marla is pointing to when she uses the term "over-riding." Someone else may feel that the character of the materials is being neglected in this description but I doubt that Marla is debating that.


R. John Howe

Posted by Sue Zimmerman on 06-05-2003 09:14 PM:

Hi John,

I am aware of the process of weaving. I know you feel rather uncomfortable with your structural analysis skills so if there is anything in my first two posts in this thread which you don't understand, please let me know and I will try to put it another way to clarify. I offer the same to anybody else who might not understand what I've said. In my last post I was not thinking of ancient weavings, I was thinking about the one which is the subject of this thread. I am hoping that Daniel or Michael can shed some more light, that is what I will be waiting for. Sue

Posted by Marla Mallett on 06-05-2003 09:53 PM:

Dear Sue,

No, as John suggested, that’s not the debate. But of course you are correct about the importance of the materials used in the development of new textile structures and traditions.

So let’s consider what those factors would be with soumak pieces, particularly cuvals, heybe, khorjin, and mafrash. Since the warps in the case of extant pieces are pretty consistently the same, we can set aside a discussion of those, as nearly all have been adequate. That leaves wrapping yarns and ground wefts.

What are the textile qualities we should consider? Tensile strength, loft, elasticity, density, felting or matting qualities, and resistance to abrasion, to name the most relevant in this case.

Cotton’s biggest advantage is always its superior tensile strength, but in most soumak work it’s not been used so as to utilize that quality. It mats readily, and abrades easily, thus on many old soumak bags we see severely worn cotton surfaces, while many wool areas have held up better. (Just take a look at old Anatolian kilims that have severely matted white cottons.) Cotton is truly lacking in terms of loft, and in being more dense and more unforgiving than softer, fluffier wool, it is less suited to soumak wrapping, because it doesn’t “cover” or compact as well. Thus often weavers have squeezed in more cotton rows of wrapping to cover the surface adequately. To add more ground wefts in areas of cotton wrapping than in wool areas, would be self-defeating; the deficiency would be magnified. In fact, the reason most soumak weavers have used cotton in soumak work has been for aesthetic reasons--to utilize its stark whiteness when bleached—a whiteness not matched by white wool.

One of wool’s great advantages is its loft—which is much, much greater than that of cotton. (That’s why it is used for tapestry, for brocading, for knotted pile, AND for most soumak wrapping.) Even when fairly tightly spun and plied, it spreads out and fluffs up to make a nice, fully covered surface. Its superior elasticity means it is forgiving, and so is much easier to wrap with to achieve a smooth and even surface. (Of course it also takes dyes beautifully, but that’s not a relevant point here.) The tensile strength of wool is very low, however, and if it is to be used for ground wefts that are covered, these ground weft yarns must be very thin. Their elasticity and flexibility however, mean that they can easily be laid in with enough ease to assume the sinuous path required of a ground weft that must compact well. But wools used as “ground wefts” in a “compound structure” combining wrapping and plain-weave interlacing are the weakest element in most single-wefted soumak weavings, especially in those with vertical pattern lines and thus nearly open slits between pattern areas. We often see the tiny wefts broken in these areas on old weavings. They are weak enough that they often break if a stiff soumak panel is folded carelessly.

Thus many soumak weavings are a compromise at best in terms of their adaptation of materials to structure. Double-wefted soumak bags (as with many coarse Kurdish pieces) are far superior structurally to most Caucasian and Persian Shahsevan pieces. And at least theoretically, one could even argue that Kurdish weftless soumak pieces are superior structurally to the far more numerous Shahsevan pieces: When ground wefts were not used, motifs and layouts were designed to not require them! Of course that meant that design limitations were severe.


Posted by Michael_Wendorf on 06-05-2003 10:26 PM:


Dear Marla, Sue, John, Daniel and all:

Sue: I am not certain I understand what you mean about the nature of cotton. Are you saying that cotton is inherently or by its nature less strong than wool? If so, I am not sure that this is correct.

John: I am sorry, but I do not understand what you are saying. I think I understand the character of both the ancient Neolithic Anatolian fragments from Catal Huyuk and the weftless soumaks woven by Kurds in Eastern Anatolia as well as what the weavers did when they made them. Exactly what is it then that you would have me grapple with, even if in a beginning way?

Marla: I am aware of and understand the distinction you make between the Kurdish weftless soumak, which are patterned and sectional or woven in one direction then floated backwards to begin wrapping the next short row, and the Neolithic fragments, which are unpatterned, widely spaced and wrapped sequentially. Your point about the particular weaving process used to make these Neolithic fabrics is valid on its own and understood. It does not require nor is it strengthened, in my opinion, by comparisons to a Caucasian mafrash panel. It is easy enough to state as you have, that there is little or only superficial parallels between the Kurdish weftless soumak and the Neolithic separated as they are by some much time, different materials and loom technologies. In any event, I doubt you need to have grown flax, harvested and retted the bast, built a warp weighted or other primitive loom and attempted to weave using these things to realize the importance of the process. Neither am I certain that even having done so, you would be able to any better affirm or deny the evolution of the technique or structure over thousands of years. Likewise, this fails to explain why Kurdish weavers used the structure even though they had looms and materials that did not dictate it. Nor does it explain how they came to use this process.

I do not know why you insist the one fragment that was illustrated in Anatolian Studies was a "NET." Burnham described the weftless fragments that were excavated as "net like" and wrote that they were used to swathe the burial bundles prior to the bundles being tied with a cord prior to internment. I guess it makes it easier to dismiss the fact that Kurdish weavers alone wove weavings in weftless soumak; not for effect, convenience or by omission, but as complete and functional weavings.

Well, now I have said too much for sure. Better smoke a Montechristo.

Best, Michael

Posted by Marla Mallett on 06-05-2003 11:59 PM:

Dear Michael,
I said nothing about loom technology, separation of time (8000 years in this case!) or even materials, in the case of your speculations about relationships between Neolithic and present-day Kurdish weftless soumak. The technical variations I've discussed are virtually the same, whatever the loom type. Shed-making options are largely irrelevant to the variations in the soumak wrapping processes themselves. I must confess, though, that I become impatient when people invent theories involving structures and processes or design/structure relationships when they have never bothered to try out the processes--in even the simplest form--for themselves. That is SO EASY to do, and would seem like a basic requirement when one wishes to present a valid academic theory. I once had an argument with a well-known rug writer (whom most of you know) over a specific tapestry process, and no matter how hard I tried to point out the error in the simple process he was describing, I could not convince him to just wrap some yarn around a picture frame and try the process for himself! He then would have discovered immediately that it was impossible to produce the structure he was describing and illustrating! He need not take my word for it! Diagrams sometimes just don't do the trick! They can be deceptive or just plain erroneous.. Sorry, but this is a frustration that I have great difficulty overcoming! That is exactly why I finally put some directions on my website for building a frame loom for simple technical experiments.

I would ask the following of anyone hypothesizing on soumak structural or technical evolution: Are you aware of the several significant variations among Caucasian and North Persian soumak weaving processes? The several significant variations in the structures? Do you know the practical reasons for these variations? Can you sort out pieces that were woven from the front from those woven from the back? Do you know why in some cases Shahsesvan soumak yarns are continuous from side to side, merely floating between design areas in which they are used, and why in other cases they reverse continually? Can you account for the weird irregularities in Western Persian and Khorasan Kurdish soumak weavings? Do you know why one sometimes wraps the warps by passing the wrapping yarn BETWEEN the fell of the cloth and itself--and sometimes not? Especially when there are reasons other than for producing countered or uncountered surfaces? For anyone who has tried out the techniques, none of these need remain merely questions. Having tried them, one would also be much closer to appreciating the significance of differences in the processes and the diverse ways that designing is affected. Of course we can only SPECULATE on the development and evolution of techniques, but I believe that for any such speculation to have much meaning, we need a firm understanding of all of the relevant variations and possibilities.


Posted by R. John Howe on 06-06-2003 11:08 AM:

Hi Michael -

You may well have sorted out intellectually the distinctions involved in these two weaves. You are a good and careful student.

I intended no stern challenge but rather wanted to make a gentle suggestion that Marla has explicated well above. We often LOOK and LOOK and ANALYZE what we see, when in fact some of the distinctions entailed are seen only when we try to DO what the weaver had to do to make a given structure.

I attended a TM rug morning program a few years ago, given by David Fraser, the current chairman of the TM board of directors. As you know, Fraser is a collector of weft twining and the author of perhaps the standard work on it.

He had beautiful and varied examples, but the thing that impressed me most about his presentation is that he had some demonstration jigs ("warps" of thin pieces of wood held in place at one end but open at the other) on which he demonstrated with color weft yarns, how the various kinds of weft twining is actually made. It was clear that part of his expertise is that he knew pricisely, and experientially, what the weaver had done while making all the variations of this rather primitive weave. Very often there are surprises and things that simply cannot be learned in other ways.

That was all I was saying. Get or make a little frame loom and try to replicate what the weaver did yourself and see if you don't learn things.


R. John Howe

Posted by Marla Mallett on 06-06-2003 02:50 PM:

Dear Michael,

Since you mentioned loom technology and materials in your last post, and said that you “doubt you need to have grown flax, harvested and retted the bast, built a warp weighted or other primitive loom and attempted to weave using these things to realize the importance of the process,” perhaps you might answer a few specific questions dealing with practical weaving issues, insofar as they are involved in the parallels you have drawn between the Neolithic textiles and the much more complex modern Kurdish weaving—both in your ICOC lecture and here:

Is it your opinion that the Central Anatolian Catal Huyuk soumak textiles were produced on warp-weighted looms? If so, how would that loom type, as opposed to a rigid frame, have affected the kind of fabric produced (discounting the presence or absence of a shed-making device, as that refinement presumably developed much later)? In what important ways would that textile construction process differ from, relate to, or account for modern Kurdish weftless soumak production? What are the important differences between producing such a fabric in a bast fiber and producing a similar structure in wool? Can the former use of bast fibers and a switch to wool account for particular features in modern Kurdish weftless-soumak textiles? What significant similarities have you seen in the ways that modern Kurdish weavers have produced weftless soumak and other soumak products? Do such technical similarities, if there are any, have importance in separating these other Kurdish soumak weaving traditions from those of the Shahsevan weavers of North Iran and the Caucasus? In other words, are they weaving traditions with completely separate evolutionary developments?

Another issue: Have you, or has anyone else, tracked down current-day weavers still using the Kurdish unique weftless-soumak processes, and observed them at first hand? Occasionally a weftless-soumak bag pops up that seems to be quite recent. What other kinds of textile processes have been used by these people and their older family members?


Posted by Deschuyteneer Daniel on 06-08-2003 05:59 AM:


We all, are here to learn and the questions raised in your postings are certainly extremely important. Unfortunately there isn't anybody on this board (or other one) who is able to answer to them. So if you don't answer to them we will miss a lot.

If we let actually on the side the "Neolithic connection" suggested by Michael you raised the following points:

What are the significant variations among Caucasian and North Persian weaving process? Is it possible to observe these variations in finished items and to use such variations to sort weavings from these area

What are the significant variations among Caucasian and North Persian and the practical reasons for these variations?

How to sort out pieces that were woven from the front from those woven from the back? Is it of any interest to sort pieces?

Why in some cases Shahsesvan soumak yarns are continuous from side to side, merely floating between design areas in which they are used, and why in other cases they reverse continually?

What are the weird irregularities you have observed in Western Persian and Khorasan Kurdish soumak weavings?

Why one sometimes wraps the warps by passing the wrapping yarn BETWEEN the fell of the cloth and itself--and sometimes not? Especially when there are reasons other than for producing countered or uncountered surfaces?

Curiously nobody reacted when I told in my essay that weftless soumak weaving, finally a very simple technique, was used in Eastern Anatolia "AND" NW Persia. Until now it was considered, on this board, that it was a "specific" Eastern Anatolian technique.

Thanks for you help

Posted by Marla Mallett on 06-08-2003 04:43 PM:

Hi Daniel,

OK, I’m in trouble, now, huh? I raised the questions above to illustrate how little most folks know about soumak processes, and to demonstrate that the issue is more complex than most assume. Once getting beyond simplistic diagrams and noting whether the work is 4/2 wrapping, 2/1 wrapping, reverse soumak, knotted wrapping, diagonal wrapping, weftless soumak, countered, uncountered, etc. most people have given the structures and processes very little thought. The fascinating weftless soumak Kurdish bags that have so captivated Michael are NOT simple things. The processes involved in them are complex, and the design restrictions they pose are also complex. THAT was the main point I was trying to make—that a relationship between them and the very simple Neolithic example from Central Anatolia is superficial and almost non-existent. ANYONE can figure out how to do that simple Neolithic variety in about 30 seconds, but I’d guess that just about everybody here would find duplicating a moderately complex design band on one of those modern Kurdish pieces quite difficult! I guarantee that to devise a design of one’s own that would be successful in the structure would prove still more challenging. To go through each of the points that I raised and that you’ve mentioned above, would require lots of magnified illustrations and pages of explanations. I’ve touched on a few of these points in WOVEN STRUCTURES (maybe just in the 2nd edition…I’m not sure, as I can’t find a copy of the 1st!), but decided that a discussion of the more complex issues was inappropriate in that introductory book. Most importantly, though, all of these issues can be understood by trying to reproduce each of the structural variations for one’s self. That’s a sure-fire way to grasp their importance. It’s because apparently no one has done such a thing that there’s nothing elsewhere in the rug literature concerning the points I raised. And yes, of course the variations should be of major help in sorting out the products of different tribal groups. For someone looking for a project, I recommend it!


Posted by Wendel_Swan on 06-08-2003 04:48 PM:

Dear Michael, Daniel and all,

Although Michael has correctly cited John Wertime's Sumak Bags about my cruciform medallion sumak bag, John seems to have made a rare error in his technical reference to it. That bag, #74 in the book, does not have extra rows of wrapping between the ground wefts (I'll refer to it as "wrapping packing."). One of the distinguishing features about the piece is the regularity of its weave. It is all wool, with no cotton highlights as seems to be common in bags where wrapping packing is found.

While the weaver used some different wrapping techniques, wrapping packing was not among them.

John cannot recall now why he made that reference, but I checked the bag again. I believe, and John confirms that this may have been the case, that he was referring to plate 48 (also mine), not 74. Plates 48 - 51 are all from the Hashtrud - Mianeh area and cotton wrapping packing is a rather common feature in this rare and finely woven group of mafrash.

Other examples in the group can be seen as plates 71 - 73 in Tanavoli's Shahsavan.

Among the several pieces in this group with cotton wrapping packing that I own is this mafrash end panel:

In the packed areas, the cotton sumak wrapping is around 38 per vertical inch, as opposed to around 30 for the dyed wool wrapping. Silk ground wefts are employed to achieve the extraordinary fineness.

The technique has created a contrast in texture and color between the undyed cotton and the wool pattern. I can imagine no structural reason for choosing to weave in this manner, but the result today is that the dyed yarns stand out boldly against the white cotton and the texture of the packed cotton wefts seems almost flat. I have to assume that this was the intention of the weavers of these sophisticated (and probably expensive when created) panels.

For those interested, the sumak wrapping is "regular" and not countered.

Many people assume that sumak wrapping is pretty much alike, but my observation has been that there is at least as much variation in the sumak techniques as there is in pile weaving. That "proto-sumak" and "weftless sumak" and "wrapping packing" all exist in NWP doesn’t surprise me at all. I suspect that there are even more varieties that we have not yet observed.

As to Daniel's comment about weftless sumak appearing in NWP bags, I will try to locate an image of a local complete khorjin that has sumak wrapping for the "bridge" between the two faces. Those of us who have seen it believe it to be almost certainly NWP, but one could debate at length whether it might be the product of the Kurds, the Shahsavan or some other group.


Posted by Marla Mallett on 06-08-2003 06:01 PM:

Daniel: Sorry, I didn’t read your questions very carefully above. Yes, all of the various soumak processes leave their traces in the textiles themselves, so that with close study they can be used to differentiate between products from different groups of weavers. Some are immediately obvious; others require VERY close examination. MOST of them I’ve touched on briefly in the book—but only briefly, because I figured that just about everybody would skip right past those remarks, having no interest in such details. It was difficult to decide what was irrelevant trivia that would merely bog down readers. This may be an appropriate subject for an in-depth article or even a piece on the website WOVEN STRUCTURES UPDATE if I can find the time. I’d have to drag you into that, Daniel, as with the “End Finishes” project!

As for weftless soumak in NW Persia, I have one saddlebag from the Hoy area that displays the structure. It’s on the website. It was bought from a Kurdish family who claimed that one of their grandmothers had made it. I didn’t think to ask my friend who acquired it if he’d come across other pieces of the sort in the area.

Best wishes,

Posted by Deschuyteneer Daniel on 06-11-2003 02:42 PM:

Dear all,
I am very pleased to share with you Marla's thougts about the fragments she received. Her conclusions are not very far from what had been observed in Wendell's bagface.

Dear Daniel,

I received your soumak fragments today and so here are my initial thoughts about the peculiarities they display.

This has to be soumak on the finest scale I've ever seen; a true tour de force weaving! Thus the fact that ground wefts were used only after every 2nd row of wrapping throughout most of the piece is not surprising. The wrapping yarns are nearly as thin as the ground wefts, and it would have been very difficult to squeeze in more wefts. The motifs are executed quite precisely because these PAIRS of wrapping rows form the majority of the diagonal design steps.

You are right: we could go crazy looking for the ground wefts, as they are nearly impossible to see. But if we look at the front side and roll the fabric--folding it weft-wise--the rows of wrapping separate quite clearly wherever there is a weft. Then we can see where there are 2 consecutive rows of wrapping, where there are 3, or 4, and can follow them along horizontally and easily see where the structure changes. The number of rows changes most often where there is a horizontal color change, but not always.

The variations seem to me quite clearly for the purpose of equalizing differences in yarn size, and solving problems related to that unevenness. It's quite possible that different people spun the differently colored yarns. Most of the reds are thinner than the blues, and so extra red rows were often crowded in--3 rows of red wrapping used alongside 2 rows of blue. But the red is also unevenly spun; in some places these yarns are thinner than in others. Because of this unevenness, the weaver has sometimes inserted short lengths of discontinuous red wrappings. The difficulties of inserting small hooked design motifs into a plain background area shows up near the tops of each figure, where the rows of wrapping waver. Here, the weaver has inserted extra discontinuous wrapping yarns as needed to straighten out the work.

In areas where the red and blue wools are used next to white cottons, still more irregularities occur. The white cotton yarns are heavier and packed so tightly to cover the surface that the weave buckles and bulges in some places. This weaver struggled with those white cottons, especially in narrow pattern parts where she needed to reverse the wrappings continually.

Yet in the design bands where the design combines a peach color and red, no irregularities seem to occur, the weave is smooth, and two consecutive rows of wrapping are used consistently. Since these fragments are very narrow, I can't guess what might have occurred in adjacent areas, where other colors might have been introduced.

Many thanks for sending these pieces! They have been fun to look at, and I appreciate that you've shared them.


Posted by Seref_Ozen on 06-15-2003 02:41 PM:

I do not have a record of the structural details of this piece since I sold it last year but I believe the posted piece and this very piece are sharing the same origin. It is extremely fine and it has great details.

best regards

Note: Seref sent this to me by e-mail, offering to remain anonymous in order not to appear to be promoting his business. We appreciate his consideration, but since he no longer owns this piece, posting it is well within our range. Thank you, Seref. Steve Price

Copyright © Jelsoft Enterprises Limited 2000, 2001.