Home > Turkmen Rugs >Uncovering the kepse gol origin
Sat, Nov 9th, 2013 08:02:26 AM
Topic: Uncovering the kepse gol origin

Unique western Anatolian rug, circa late mid-16th century, with archetypal turreted-outline ‘gol’ similar to that found on typical kepse gol; Collection TIEM, Istanbul

The kepse gol stands alone; there is no other Turkmen gol even remotely like it.

This is enough to provoke questions about its origin.

The earliest form of a recognizable kepse gol is the unique double-gol we described and pictured in our paper “An Enigmatic Main Carpet: The Ballard MC”.

However, the carpet in which it appears is not, in our estimation, the earliest with a kepse-type gol.

That honor goes to the ex-Ballard Collection multi-gol MC, now in the Metropolitan Museum of Art in NYC thanks to James Ballard’s bequest.

Possible kepse gol archetype; ex-Ballard Collection MC, Metropolitan Museum of Art Collection

Unfortunately the Ballard carpet is cut in half and the gol above, and others, are disfigured.

This sad fact does not diminish calling it a fascinatingly curious weaving. Nor does it prevent viewing a stellar expression of the iconography archaic Turkmen weaving culture embodied.

Upper: Archetypal kepse gol, cut-Ballard multi-gol MC; Lower: Prototype kepse gol; multi-gol carpet sold at rippon-boswell 2009

It is truly unfortunate for Turkmen rug studies so few archetype weavings remain.

This is, in our opinion, not the result of the ridiculous wive’s tale assertion “The Turkmen just mindlessly consumed their rugs, treating them as everyday objects to be used and then discarded.”

Yes, this probably was the case for certain domestic weavings but for ones with complex iconography this surely was not the case, and to believe so is blatantly ignorant.

No culture anywhere in the world treats objects with proprietary sacred symbols, their spiritual imagery, haphazardly.

They revere, honor and protect items -- be they woven, sculpted, painted or carved – which display these insignia.

They will even kill and go to war over them.

So it is completely absurd to believe the Turkmen just carelessly used and abused weavings decorated with their most sacred iconography.

It is even more incorrect to believe certain motifs on Turkmen carpets had no spiritual content or connection.

Two possible transitional versions from archetypal kepse gol to the prototypes we have illustrated; Upper: ex- Ballard multi-gol MC; Lower: later ex-Hecksher/deYoung Museum MC(2)

Again, it is highly unfortunate no actual proof of this exists directly from Turkmen sources, or indirectly reported by outsiders who might have witnessed first-hand certain ceremony or rituals.

Therefore Turkmen studies must take for granted, and as fact, some pre-commercial period weavings possess spiritual content, and their iconography was often in part or wholly non-secular.

Now then, which motif are sacred icons and which are secular design/decoration?

This is an equally difficult question to answer but two type of indicative evidence are worth consideration.

First: What icons are primarily seen on early weavings and rarely on later ones?

Second: What icons are seen in the weavings of only one, or two, Turkmen groups; and what are seen on many?

This paper is not going to explore this important topic but rather mention it as background to try and understand the history of the kepse gol.

Before beginning this discussion one apparent, but rarely discussed caveat needs to be noted: Before 1800 very few complex patterned Turkmen weavings were produced; and before 1700 far, far fewer. The same goes for before 1600.


As these centuries passed there was an ever increasing urbanization of Turkmen groups. The available historic data shows this was the case enumerating increased number of yurts and livestock herds possessed by each group living in these areas.

Along with this urbanization (which should not be considered in the contemporary sense of moving into cities but settling into oases, villages and towns rather than remaining nomadic or semi-nomadic) came population increase as well as increased material wealth.

And one important type of wealth was carpets, trappings and woven goods.

The historical record also supports these conclusions, as do the numbers of extant Turkmen weavings from each succeeding time-period.

Nor is it coincidence the Turkmen measured material wealth by livestock herd ownership, or that it was these animals that produced the raw material necessary to create their weavings.

Those weavings were also considered important material wealth, besides for their (suspected) hereditary, heraldic and spiritual value.

In brief the way we see this demographic was larger and larger numbers of Turkmen weavings were produced with each passing decade, and century, to satisfy the now far more numerous and wealthy Turkmen clans.

Forget about export market possibilities which likewise increased exponentially as the 19th century dawned and then progressed.

We believe concomitant with this increasing supply of weaving came a decreasing number purposely created and imbued with sacred iconography, as the stresses of societal change, ie urbanization, the spread of Islam, and the internal conquests of weaker Turkmen groups by more powerful ones destroyed direct connection with the historic roots and sacred nature of the weaving culture and its proscribed proprietary iconography.

This seems apparent from the increased supply of secular weavings displaying some formerly, and most probably, non-secular symbols. These leaked into the secular domain due to the persistent breakdown of the sacrosanct proprietary nature, tenet and prohibition followed by the ancient Turkmen weaving culture.

The ostensible reason for this, and one also supported by the historical record, is the ever-decreasing number of individual Turkmen groups, though the overall numbers of Turkmen people were increasing.

Many minor groups virtually disappeared without a trace, while others became amalgamated with more powerful ones. Raiding and killing among Turkmen groups decreased, another factor causing increased overall population.

It is believed these ‘disappeared’ groups, and others known to have been absorbed, had their proprietary weaving iconographies absorbed as well.

This was no doubt one of, if not the, main reason for formerly non-secular, sacred, symbols falling into secular usage.

What could be more insulting than taking your enemies prized sacred icons and placing them on your floor coverings and domestic products destined for hard daily use?

Again, the apparent large number of certain types of 19th century pre-commercial period weaving, like some engsi and the Tauk Nauska gol main carpets, appears to be the result of this development.

We cite the huge number of extant Tekke engsi as perhaps a prime example. Also, the numerous examples of Yomud, Chodor, Ersari, Kizil Ayak, and their still poorly understood sub-groups, who produced an inordinate number of Tauk Nauska gol MCs.

This seems to explain why there are so very few pre-1800 engsi and Tauk Nauska MC while there are so many 19th century ones.

Though this lens it is not difficult to see how a formerly sacred-protective format like the engsi, and its very specific iconography, became desirable as domestic door ‘furniture’ for the enlarged numbers of Tekke clan, the most powerful, dominant and numerous Turkmen group throughout the 19th century.

Or how certain proprietary and individualized versions of the Tauk Nauska gol, produced by Yomud, Chodor and eastern Turkmen groups, became homogenized into the multitude of nondescript Tauk Nauska gol carpets made during the second half of the 19th century.

OK you might ask where is all this leading, and what does it have to do with the kepse gol?

RK briefly discussed the kepse gol in our “An enigmatic Main Carpet: the ex-Ballard MC”, noting its earliest appearance is exclusively in multi-gol carpets.

That paper now appears in the Best of RugKazbah Topic Area and we are sure many readers have already seen it.


Detail of a prototype kepse gol we call skeletal

Here is the complete MC where this gol appears

Multi-gol carpet sold at rippon-boswell in 2009

This carpet appears to be circa 1750; maybe slightly earlier, though not pre-1700.

But the far more simplistic iconography, compared to the typical kepse gol, and skeletal format are key signs it is different.

We view these factors implying it is significantly earlier than the typical form, probably one of its prototypes.

Also the fact its skeletal format is unique, and the carpet in which it appears undoubtedly pre-1800, add considerable weight to our position.

The basic form of the kepse gol is there, one that clearly underwent significant accretion in its transformation to the later, typical, well-known, format.

The process that took this prototype, and others we will illustrate, and transformed them into the typical kepse gol is the theme of this exercise.

We have always considered the typical kepse gol a contrived iconographic element, one lacking genuine historic roots in Turkmen weaving culture. And that assumption proves correct according to the chronology we have developed.

Put another way: Until publication of this paper, and some ideas in our forerunner “An Enigmatic main carpet: The Ballard MC”, no one in Turkmen studies recognized what we see as strong evidence of how the typical kepse gol originated.

Our theory rest on the premise the gol displayed in the enigmatic Ballard MC represents the only existing version of the archetypal kepse gol, which then over time became transformed into the prototypes, like the skeletal and other versions below.

Upper: Archetypal kepse gol, cut-Ballard multi-gol MC; Lower: Prototype kepse gol; multi-gol carpet sold at rippon-boswell 2009

One small piece of evidence to support our concept can be found on another multi-gol MC.

Detail multi-gol MC with transitional archetype to prototype kepse-gol (number 2 in this photo); illustrated in “Turkmen”; Washington Textile Museum, figure 42; Wher Collection

This ‘look’, the thicker gol segment on the left side, echoes the drawing of the archetypal kepse gol, and shows the initial codification that transformed it into a typical kepse gol.

Actually this is only an illusion created by color juxtaposition, as there is no appreciable height difference left to right.

Whether or not this illusion was intended cannot be known, but the illusion is there.

Also worth note are the pincer-like finials at the left and right horizontal terminus of the gol.

Again the left side is accentuated by color choice, adding to the idea the illusion was purposely done.

This unusual, pincer-like motif is rarely seen and will be mentioned in the description of the next MC.

Recently photos of another multi-gol carpet, with a prototype kepse gol, were emailed to us and not surprisingly it provides additional support for our theory.

This multi-gol MC can be considered somewhat a game changer as it displays four distinctly different gol.

When we received the pictures the carpet was unknown and unpublished, but since then it has been consigned to a European auction and now has been published.

We have placed numbers next to the four gol in the detail below to facilitate discussion.

Number 2 is the prototype kepse gol.

We believe this kepse gol to be a contemporary prototype similar, but actually quite different when carefully examined, to the skeletal prototype.

These are not the only MC with versions we consider to be prototype kepse gol.

Here are two others

Upper: Multi-gol MC that appeared at auction in Paris, 2012; Lower: Multi-gol MC, deYoung Museum ex-Hecksher Collection(1)

And the complete Wher collection MC

While ostensibly similar they all have subtle differences, and these can be used for comparative dating as well as help to hypothesize how the kepse gol developed.

Note: Because there are two ex-Hecksher collection multi-gol MC we have numbered them (1) and (2) to avoid confusion.

At this point it is probably pertinent to comparatively date these rugs on a continuum.

Chronologically the vertically cut Ballard MC is the oldest; next would be the ex-Hecksher/deYoung MC(1); then the other ex-Hecksher/deYoung MV(2); then the multi-gol MC soon to be auctioned; next the rug which was auctioned in Paris; next the Wher collection (black and white photo); and last the rug auctioned at rippon-boswell (the skeletal kepse gol).

Perhaps one day an even earlier than the cut Ballard MC archetype kepse gol will be discovered.

Maybe it will even not appear in multi-gol format.

Frankly we are dubious this will occur but anything is possible as the boundaries of Turkmen studies have consistently proven quite elastic.

In conclusion we offer the following:

1. The enigmatic Ballard multi-gol rug has no kepse gol per se. It does, however, have a gol we consider to be the earliest presently known version, one that shows the probable archetype.

Two possible transitional versions from archetypal kepse gol to the prototypes we have illustrated; Upper: ex- Ballard multi-gol MC; Lower: later ex-Hecksher/deYoung Museum MC(2)

2. Were the Ballard rug complete, and not cut in half, there might be some additional clues how that actual archetypal kepse gol might have appeared.

But it isn’t.

All we have to rely on the later ex-Hecksher/deYoung MC’s(2) version, which is early enough to support our theory it and the ex-Ballard collection display the best indications how an archetypal kepse gol probably looked.

3. Careful study reveals these two unique gol possess just too many typical kepse gol iconographic elements, perhaps the most significant an interior horizontal central axis, inner (roughly shaped) hexagonal central medallion, and the unusual jagged, stepped, outer perimeter (more about this soon) to be coincidence.

4. To try and pinpoint how the rest of the kepse gol elements came together requires insightful comparative analysis, and some imagination.

The historical record shows as time progressed the number of proprietary Turkmen gol (those used on only one groups weavings) decreased and the number of shared gol (those used on more than one group’s weavings) increased.

It is also implies some of those that remained active, like the kepse gol, more than likely incorporate features formerly belonging to those disappeared gol.

5. This process can now be far better understood thanks to the past 25 year publishing avalanche of newly discovered rare and early Turkmen weavings.

It also helps locate where several key kepse gol iconographic elements might well have originated. But how, by whom or when they coalesced into the standard full-blown version is something it cannot explain.

Though problematic these unknown do not diminish the possibilities cited in this paper.

6. The first of those iconographic element is the anchor motif every typical kepse gol displays.

Right: Kepse gol from a Yomud main carpet we date earlier than most, circa 1825; Left: detail of its anchor motif

A proto-anchor even appears in one of the oldest prototypes.

Arrows pointing to kepse gol anchor motif; Soon to be auctioned multi-gol MC

7. After considerable research we believe to have located where the anchor originated.

Some years ago RK discovered a Tekke large format torba (LFT) that is remarkable, and without doubt the earliest Turkmen weaving we know.

We have no qualm dating it circa 1600 or earlier.

We have listened to suggestions for submitting it for c14 dating. But since we do not trust c14’s reliability to date a weaving which has been subjected to many types of contaminants through centuries of use we haven’t, and have no plans to do so.

Detail archetype Tekke LFT gol; previously published RugKazbah.com; RK collection

8. For those who can’t see it, here is a detail of the anchor, and one from a typical kepse gol for comparison

Left: Detail Archetype Tekke torba gol anchor icon; Right: Detail for comparison of a recently published Chodor MC with typical kepse gol anchor motif

9. The bird/animal head icon in the Tekke LFT gol anchor does not appear in typical kepse gol.

However it does appear in some of the earlier prototypes(see below).

This is not surprising. We can point to other instances of early gol iconography where this significant icon appears but then is omitted, ie forgotten, in later copies.

10. While the archaic Tekke LFT gol and kepse gol comparison might appear far-fetched, there is some corroborating evidence – a set of factors these two quite disparate gol share.

RK has mentioned the set concept before – three or more shared iconographic elements that identify relationship between seemingly unrelated weavings.

10. Within the center hexagonal medallion of almost all the prototype kepse gol is a particular type of (equilateral) cross with a circle, or kotchak, at each terminus.

Prototype kepse gol with crosses in their central medallions; Left: detail ex-Hecksher/deYoung multi-gol MC(1); Right: detail, soon to be auctioned multi-gol MC

This same cross appears in the central medallion of the archaic Tekke torba gol.

Right: Archaic Tekke gol; Left: blacked-out detail of the same

And the Tekke LFT’s (roughly) hexagonal outlined in blue medallion is another comparative worth noting.

11. The third part of the set is the bird or animal head icon.

Left: Archaic Tekke LFT gol animal-head icon; Right: a prototype kepse gol animal head icon, ex-Hecksher/deYoung Museum(1)

Again, unsurprisingly, it also appears in the archetypal ex-Hecksher/deYoung multi-gol MC(2) kepse gol

Bird/animal head icon embedded in the archetypal kepse gol, ex-Hechsker/deYoung Collection multi-gol MC(2)

Additionally, the typical kepse gol turret's jagged peaks above each anchor are reminiscent of the pair of grander orange and white ones above the Tekke LFT anchor, as well as the halo of them attached to the hexagonal outer medallion outlined in blue that surrounds the inner octagonal medallion with the cross.

12. We sincerely doubt sharing this set is coincidence.

At the least it shows tangible relationship, and at the most the transcription of important iconography from an outwardly unrelated archetype model to later subsequent copies.

13. We trust it should not be hard to reason how the animal/bird head anchor icon from the archetype Tekk LFT gol became integrated and embedded, albeit it in a far less defined form sans bird/animal head, in the synthesized later typical kepse gol.

14. Now don’t get us wrong there is a long (age) gap between the archetype Tekke LFT and any main carpet with a typical kepse gol, with a transformation/transition process that cannot possibly be fully plotted.

We likewise realize our kepse gol speculation is just that, speculation. But the documentary evidence we offer provides a realistic roadmap.

15. Then there is the simple fact no early, ie pre-1800, carpet with a typical kepse gol exists and from that period only multi-gol MC with prototypic kepse gol are known.

This goes a long way to support our contention the typical kepse gol developed over time, and its use in carpets with a major/minor gol format a rather late addition to the Turkmen gol vocabulary.

16. Lastly, the tantalizing, and relevant as we see it, detail of an early Anatolian carpet published at the beginning of this paper deserves mention.

There are strong links between certain types of very early Anatolian carpets and those of the Turkmen. This is due to the well-recognized fact Turkmen groups colonized areas of Anatolia from the 13th to the 16th century, bringing their traditional weaving vocabularies and iconography.

Here is a photo of the carpet that detail comes from

Collection TIEM, Istanbul

We referred to this carpet as “tantalizing” because of its connections to the kepse gol.

Even a cursory inspection reveals this carpet has a two-gol layout. The first being the one we illustrated, and do again below on the left. The second is on the right.

Details of the two different gol from the Türk ve İslam Eserleri Müzesi (TIEM) carpet

There are a number of very interesting iconographic features the gol on the right also shares with the typical kepse gol, as well as its prototypes.

Although the turrets are not nearly as pronounced, they are still present.

More significant is the shape of the outer (red) hexagonal central medallion that mirror the typical kepse gol outline.

And while these pictures are unfortunately low resolution they are good enough to see these features.

Those pincer-like elements at its horizontal ends also appear on the multi-gol carpet that is to be soon auctioned (gol numbers 2 and 4 below).

Seeing these on a “C” gol, number 4, is unusual, perhaps one other indication this carpet is not as early as the growing gossip surrounding it believes.

The far more regular and simple fork at the end of all the typical kepse gol no doubt a vestigeal remain.

While hypothesizing these and other possibilities let’s get even father out and suggest the kepse gol archetype was a Timurid invention.

Clearly, the chances for it being Anatolian and then going back to Turkmenistan are far less than it originating in a Turkic weaving tradition, like that of the Timurid, or one of the earliest Turkmen groups who migrated into Anatolia.

Such an issue and others raised in our examination of the kepse gol, are at the cutting edge of rug studies.

Perhaps the coming decade will provide new discoveries of both carpets and ethnography which will help to answer these questions.

And, we are sure, raise others.

Home   Buy/Sell at the Kazbah   Terms Of Service