Posted by R. John Howe on 03-29-2003 08:20 PM:

15th Century Andalusian Rug

Dear folks -

Carol Bier, formerly the Curator for Eastern Hemisphere at The Textile Museum and now, a Museum Associate, has curated two interesting exhibitions that opened recently.

The first of these exhibitions features "Andalusian Rugs," a term that refers to Spanish rugs woven under the period of Muslim rule and Mamluk rugs woven in Egypt. These pieces date to the 15th and the 16th century.

This post focuses on one piece in the Andalusian exhibition.

This 15th century rug has never, I think, been exhibited before, perhaps in part because it is in very poor condition. Spanish pile rugs, many of you will know, are woven with particular knot that encircles only one warp. They are always offset knotted from row to row and this carpet is so worn that its structure shows clearly over most of its surface.

Carol Bier is interested in thoughts about this piece and arranged for The Museum to send me an image of it. Unfortunately, that image is very dark and to make the features of this design more visible, Filiberto has been playing with it a bit backstage at my request. So the colors in these images can no longer be regarded to approximate those of the actual piece, but it is true that the piece viewed face to face is not as dark as the image the TM provided.

This is an interesting rug because it has field devices that have features much like those of some Turkmen weavings. There is a main gul with some Turkmen usages and a minor ornament that functions like a minor gul but which seems European, as do the borders. Here is a somewhat closer look at the "major gul" device.

Notice that at the top and the bottom of this piece there are features very similar to those of Holbein usages that also appear in some Turkmen rugs. And there is a diagonal use of color. The odd thing is that the sides of this major field device are flat.

So what do you think? Has anyone encountered a similar piece? What do you think of the seeming similarities to Turkmen designs? And what other thoughts or conjectures does this piece raise for you?

Thanks in advance for anything folks are willing to contribute.


R. John Howe

Posted by R. John Howe on 03-30-2003 08:18 PM:

Correction and Attempt to Fix TM Link

Dear folks -

I wrote Carol Bier and indicated that I had put the piece above up and she looked in on it, then sent me a reply saying in part that

"...The TM link that you posted went nowhere for me. Also, the description of the two exhibitions got a little garbled; the Carpets of Andalusia relates to both 15th and 16th century carpets; the Mamluk only to 15th century."

Thanks, Carol, for the correction.

Here's another attempt to provide the link from the TM site.

I'll give the general TM link first. With this one you will have to look about for the Andalusian and Mamluk links.

And here's another attempt to "drill down" a bit to provide the current exhibitions link. Scroll down for information on the Andalusian and Mamluk exhibitions.

Hope that helps a bit.


R. John Howe

Posted by Vincent Keers on 04-03-2003 03:13 PM:

The missing link

Hi John,

Found the missing link between your penny rug, the Andalusian rug and the Turkmen güll.

Some of the design in this 10'th century textile seems related to the Andalusian Rug.
It was found in the Pyrenean Mountains between France and Spain.
At the time this was made, in the Spanish court workshop, the Turkmen did not yet know they were Turkmen.
They thought themselves to be Uzbek or Oguz.

Best regards,

Posted by Ali R.Tuna on 04-04-2003 11:27 AM:

Interestingly , a detail of this carpet is published in Turkmen book , in Jon Thompson's article. There , the details of the Gul are more visible.

The Gul is from the same family as the Small pattern Holbein Guls . There are Small pattern Holbeins with the inside quartered that way , in Philadelphia Museum (see CGEllis) , in Budapest Museum (see batari) and probably one exhibited in the last ICOC in Italy as far as I recall.

The sides of the Gul are not straight but the "knotted band's "straight part is longer than the angled part such that it gives the straight effect So for me , the gul is a SPH overall.
Except that the knots of the knotted band are executed in a very distinct way and one can easily follow the endless knot trail.

What is more interesting to me are the double figures in the field of the Gul. These guls are usually having a central design. Here there are two motifs in the field.
Also the colour quartering of the guls echo themselves in the outer border design.

Given some recent finds that indicate the similarities between Anatolian Holbein carpets and Eastern Turkestan carpets , it might also be speculated that this rug was inspired after an eastern Turkestan original ? Just an impression.

Posted by R. John Howe on 04-06-2003 07:02 AM:

Dear Ali et al -

I am sorry not to have acknowledged your interesting post earlier (we are deep in preparations for ICOC X now).

I had not remembered that a black and white detail of this rug was in the 1980 Macke/Thompson "Turkmen" catalog. I will put an image of this detail up as soon as I can manage that.

Thanks also for pointing out that this image makes clear that while the sides of this "gul" have "flattish" areas, there are also indentations that make its shape more complex than my description of it.

You mention the center of the major "gul" and John Collins has sent me a clearer color version of both the major gul and of its center and I am hoping that he will post them and have some comment about them.

Carol Bier almost assuredly knows of the presence of this detail image in Thompson/Mackie, but did not mention it to me. I might over-read this omission a bit, but it may signal that she finds Thompson's suggestion that this design likely has a Turkish source less than convincing. She makes no reference to a possible Turkish source in her gallery labels.


R. John Howe

Posted by David R. E. Hunt on 04-06-2003 04:00 PM:

Turkic Influences

Dear R.J. Howe and All- While the structure of this rug might point in other directions, I think that the design of this carpet is quintessential Turkic.The gul format and the border derived from cufic script are of turkic origin, and the relationship to the Holbein carpets is clearly evident.

This scan of a carpet simultaneously described by both Erdmann as a " free paraphrase of the Holbein Gul" and Gantzhorn as a" Holbein Gul Prototype", to me demonstrates an appreciable of consanguinity, in both gul element and border, with the Textile Museum Carpet, as does

this divan cover from Pacific Coast Collections.
Perhaps it is no coincidence that two examples of monumental architecture, one located in Marrakesh and the other in Seville,are both designed by the same architect and possess strong elements of the Seldjuk Style.These structures, the Minaret of the Qutubiyah Mosque and the Minaret of the Great Mosque in Seville, were constructed in the the 12th cent. and attribuited to the Sevillian architect Guever.

Posted by Ali R.Tuna on 04-06-2003 04:35 PM:


I must just correct one point in my post. The carpet is published partially in Turkmen but the article in which it is embedded is "Turkmen weawings in Historical perspective" and is from Louise W.Mackie rather than J.Thompson.

She is showing it as an example of quartered motifs in Turkmen tradition. She writes " The oldest known example was woven in Spain in probable imitation of a lost Turkish model similar to SPH"s.

Since than the carpet research has made some progress but I do not know if any new element can be brought to what she says. I would also look at the similarity in quartering between the guls and the border and would not also eliinate some Eastern Turkestan flavour in colouring and patterning.

By the way , good luck for all the ICOC organization !


Posted by Vincent Keers on 04-09-2003 07:21 AM:

Ola Wappo John,

I wasn't joking. Maybe your Pennyrug was a bit out of the order, but for the rest I was deadly serious. The Andalusian rug seems to incorparate some different artistic levels.
The textile I posted is filled with all kinds of independent design. The Holbein Gull has a more continuous design. So has the Andalusian rug, but the two main Gulls are different.
The left Gull is more like the textile.
The right Gull is more like the Holbein.
The secondary Gull looks like a family arms, but I can't find which one.

Do you know where this rug was found?
Did it come from a Cathedral, monastry?

Best regards,

Posted by Sue Zimmerman on 04-11-2003 03:55 PM:

Please, someone tell me that I am not the only one, who, upon seeing these guls, said to themselves, "Ah-h the Moors!" Except for the fact that they are surrounded by those early "Renaissance" wreath-like things in the boarders and Anotolian style "brackets" holding the guls down, this rug lacks the goofy late dark ages emerging European stuff of other Spanish rugs I have seen. I don't see an Anotolian beat either. I see Africa. Am I alone in that? Where did that sophisticated outer boarder come from? Sue

Posted by David R.E. Hunt on 04-12-2003 11:34 AM:

Further Illustration

Dear Sue and All- I understand your confusion, especially considering what the adjustments to the photo mentioned By R.J Howe have done, but Turko-Arabs have represented the predominant culture in Morocco and Andalusia for centuries. True, Marrakech is predominantly black, but has not been the center of Maghribian culture for some time, and not when this rug was woven. The following scan further illucidates the Turkic origins of this rugs design- Dave

Posted by R. John Howe on 04-12-2003 10:05 PM:

Dear folks -

I am sorry not to have been able to follow this thread I started more closely.

A few thoughts.

Vincent - I see what you mean and that your post was not mostly in jest. And we did not say clearly enough that in fact the treatment of the centers in the major guls alternates between a two-device version and a single device version. Your graphic shows that clearly. A "penny rug" similarity to the single device version is probably a bit of a stretch but thank you.

About the "secondary" gul: I think it looks European and the borders seem a bit that way to me too. We sometimes forget that designs in this geographic area would not just be influenced by sources at a distance, such as a possible version of a "Holbein" device, perhaps imported from Turkey or even from Central Asia, but there were very real local design repertoires as well. (Think of such sources of influence as Roman floor mosaics, for example. A Roman legionnaire retirement community was located in Timgad in North Africa and there is still a museum there with a wide array of "European" derived mosaics.) So it would not be surprising if such rugs were mixtures.

Sue - Early in my first post in this thread there is the following sentence:

"...The first of these exhibitions features "Andalusian Rugs," a term that refers to Spanish rugs woven under the period of Muslim rule and Mamluk rugs woven in Egypt. These pieces date to the 15th and the 16th century."

It is my understanding that the "Muslims" of this sentence are the same folks as those to whom your "Moors" reference points. So that is a known given here. That's what "Andalusian" refers to. Spain during an era of Muslim domination.


R. John Howe

Posted by David R.E. Hunt on 04-13-2003 03:47 PM:

Origins of the Minor Gul?

Posted by Vincent Keers on 04-13-2003 07:25 PM:

Hi John and all,

Think I didn't see what I was thinking
No. I didn't understand what I thought I saw
Well, this is a getting more and more insane
The Andalusian rug uses one warp/one knot.
You said something about offset knots.
That's where the Spanish production sets itself apart from all other productions.
An offset knot can only be made because you work on two warps.
In Spanish production there is no need to use offset knots.
The left Andalusian gull has an edge that can be easily made with a one warp production.
With a two warp production it's difficult to make this kind of gull edge. The two/one devise version aren't the subject.
So I was thinking....Yes I was....don't know why.....
that I saw a needle treatment in the 10'th century textile, that explains different Andalusian edge treatment.
But then I saw that the second Andalusian gull had a Turkman edge too.
So...what is happening here?
I see too much? I think to much? I drink to much?
Whatever. This Andalusian rug seems to show us the transition from one kind of gull into a second kind of gull.
A one warp treatment to a two warp related treatment of the same design.

Did I make any sense?
If not, it's nice thinking about because most things that do seem to make sense, aren't nice to think about.

Best regards,

Posted by Vincent Keers on 04-13-2003 07:34 PM:


I only had
two glasses
of wine.
good Spanish wine: Berberana Dragon.
before I'm
it's related

Posted by R. John Howe on 04-14-2003 12:12 AM:

Hi Vincent -

Only time for this much without an image (Eiland and Eiland has a nice one on page 37 if someone wants to put it up):

Spanish rugs are truly offset in their knotting. That is, on a given horizontal row, a knot is tied only on alternate warps. Then two rows of wefts are inserted and the next row of knots are tied on a different set of alternate warps. And so on.

The Spanish knot can be tied several ways. The more common construction begins by laying a piece of pile wool over a warp, then taking it down on both sides and then crossing the two ends behind the warp and coming to the front through the separate loops (macrame people would say "bights") that are formed on each side.

Peter Stone shows a somewhat different construction in his "Lexicon."

The confusions possible about the things we are discussing are not necessarily the result of a little wine, but could lead one to have recourse to some.


R. John Howe

Posted by Filiberto Boncompagni on 04-14-2003 12:34 PM:

I received from John the following image and text:

While resting for a moment from this ICOC X craziness, here is the Eiland and Eiland image and text regarding the Spanish knot.

Looking at it again I see that my previous description was not quite correct and that in this version the pile knots do NOT come up inside the loops on both sides and that as a result this knot is not quite as firm on the basis of its own construction as it initially appeared to me.
The symmetric knot has some firmness of structure but can loosen through its rather slack collar.
The asymmetric knot is least firm of all and really needs to be pinched between two rows of wefts since it is composed of a half-hitch combined with an inlay.
A version of the Spanish knot that was firm on the basis of its own construction would likely be convenient for the offset character of the knotting in Spanish rugs because the knots are each a warp apart on each side and do not give one another much lateral support.
Footnote: It appears to be that Marla Mallett has not treated the Spanish knot at all in the initial edition of her book on structure. Surprised me.

John Howe

Thanks, John

Posted by Sue Zimmerman on 04-14-2003 02:47 PM:

As can be visualized from the illustration above, diagonally drawn lines in the guls would give a less "stepped" appearance and more smooth edges would result from this knotting technique. When knotting, the weaver would leave the shed slightly open so that only the warps to be knotted would be in a forward position. The warps to be left unknotted would be out of the way. In this rug adjustments for "unstepped" horizontally and vertically drawn lines in the design must be considered.

The illustration has two single wefts placed in the shed which was left open after the row of knots was completed. (?) The shed was then changed for the next row of knots. On this rug, for smooth horizontal and vertical lines, the weaver would add a knot on each outer unknotted warp shown. She would continue to do this, every other row, where these lines were needed. To keep these knots in place either the shed would have to be changed before each single was placed between the rows of knots, or, more likely and better, three rows of singles would be woven, changing the shed between each row. There are other ways, too.

Spanish knotting is very versatile. It provides ample opportunity to adjust knots to the needs of designs being woven, as they are being woven. The best of both worlds.

The illustration shows each knot composed of a single. For rugs like this one, which require straight horizontal, vertical, and diagonal lines, the best way would be to use two singles per knot. When the weaver got to a place where the design needed straight horizontal and vertical lines a single could be used on the outer two knots of each line so the same amount of yarn would be used for each row. This would happen every other knotted row, in this case, within the boundaries of the lines. This is akin to a Chinese method of getting more detail into rugs with very low knot counts. It is also very close to construction methods of some Turkmen tent bands, which are probably derivative. Sue

Posted by David R.E. Hunt on 04-15-2003 06:05 AM:

Afew quick thoughts

Hi Sue and All- I find this line of thought compelling, that which would state that the type of knotting employed in a rug might modify the appearence of the design. Also, what is the origin of the "Spanish" knot. Anyone thought of the possibility that it might proceed from the weaving techniques of the indigenous Moroccans(i.e., Berbers?) or other non Turko-Arabic North African Peoples? Also, one thing which suprized me so much about my experiences in Morocco was the realization that so much of that which I as an American attributed to the spanish, as in building materials, architectural design and ornamentation in general, are in reality Turko-Arabic .- Dave

Posted by Vincent Keers on 04-15-2003 10:31 AM:

Dear John,
Why are Spanish knots offset?
They can skip a warp, and they can knot on everywarp at the same level.
That doesn't make any difference in the outcome.
Can't see the offset in this meaning that a warp is
shared by two knots in the same vertical orientation.

Made a drawing:

Yep, looks sharp.

Without a sophisticated loom or offset knotting,
the Andalusian curvilinear gull could not be made by Turkmen.
But the Andalusian/Spanish knot could create the rectangular gull.

Where did the Spanish knot come from?
From curvilinear designed needlework that was translated into a knotting system
that could compete with it's originals?

Best regards,

Posted by R. John Howe on 04-15-2003 11:02 AM:

Hi Vincent -

I agree that the character of the "offsetting" for the knots that only circles one warp is different from that in which the knots employ two warps.

And it may be that we shouldn't be using the same term to describe these two structures.

That's why I was hoping that Marla Mallett had treated it and was surprised when I couldn't find that she had. She would be very good about suggesting language to describe these two structures both accurately and distinctively.


R. John Howe

Posted by Su Zimmerman on 04-15-2003 12:50 PM:

In the last paragraph of my above post I neglected to say that when the one ply knot was being added to what would otherwise be a vacant warp in the row, added to provide a smooth edge to a vertically designed design element, the next one ply knot over would be the ground color.

For an easy to visualize example think of a plain brick wall. At the outer edges of the wall every other course of bricks laid will end in a half brick to compensate for the construction method. Doe's it make a difference? I think so. Sue

Posted by Vincent Keers on 04-23-2003 10:18 AM:


I came across page 110 in "Chinese an Exotic rugs" by Murray.L Eiland. He describes the fragments
(4'th/5'th century) that have been found by mr. Stein at Lou-lan station in the Lap Nor lake region.

"The knotting, as determined by subsequent examination, sets these pieces apart from both the earlier
Pazyryk finds and the modern carpets of central Asia. Two varieties
of single-warp knot are described by Sylvan, although other observers report only
the so-called Spanish knot, which was apparently widespread during the
first millenium A.D. etc.etc.

So, the "Spanish" knot is a "Lou-lan" knot or a "Sinkiang" knot it seems,
and I think it's the best knot to copy textile related curvilinear designs.
So for me the Andalusian rug shows the best of two worlds.

Best regards,

Posted by Sue Zimmerman on 04-23-2003 09:33 PM:

Will someone in DC take a few moments to rescue this rug before they fold it up again? How much more can it take? Sue

Posted by Vincent Keers on 04-24-2003 07:48 AM:

Not yet,
Dear Sue,
and others,

I questioned the 2 different gulls and one of them is more textile related than the other one.
John posted the question about the origin of the Spanish knot.
David posted the question about Morocco.

The single warp knot has been used wherever rugs have been made.
It has been found in East Turkestan.
It has been found in Cairo as well and Copts used single warp knots.
I'm waiting for a single warp knotted Moroccan rug to pop up. It's out there.
The single warp knotted rug is less strong than two warp knotted rugs.
So maybe that's why we can't find 8'th,10'th,12'th 14'th century rugs?
But, why was the single warp knot used?
(The Pazyryk rug shows two warp knots. So this knot was available)
The need for detail?
The message behind the curved line?
It had to be done with mathematical precision in order to get the message thru?

So, from one comes another.

Best regards,