Archive for May, 2009

“Easy to Weave; Hard to Weave,” Part 1: The Lecture

Posted in Uncategorized on May 19, 2009 by rjohn

On February 28, 2009, John Howe  (that’s me)


gave a “rug morning” program at The Textile Museum here in Washington, D.C., on the topic of “Easy to Weave; Hard to Weave.”

There were two parts of this program: first a lecture, and then examination of some illustrative pieces I had brought in.  Folks in the audience had also brought quite a few pieces, illustrating their own thoughts about this topic.

Dan Walker, The Textile Museum Director, introduced me from notes the TM had asked of me.


Dan said that I had been an instructional designer in business, academia and the Federal government for over 40 years.  That I had an early interest in textiles, worked in a clothing store, during high school, had learned what a good piece of cloth was, and had taken 155 shirts with me for my first semester of college.


This is my grin, acknowledging the accuracy of this minor, but bizarre fact.

Dan suggested that this seemed like a scheme for avoiding laundering.  In fact, I was rather fastidious, then, and changed shirts at least three times a day.    🙂

Dan also noted that I am a regular attendee at these Saturday morning RTAM sessions and that I have presented, on occasion.

We both forgot to note that I also spend a bit of time sharing virtual versions of some of these sessions with a wider internet audience.

First, the lecture:

Note:  One of the advantages of a “virtual” version of something is that it can be changed if that seems appropriate.  I have sometimes made changes in this virtual version on the basis of my own initiative, but, since I quote Marla Mallett extensively below, I have had some subsequent conversations with her and some changes I have made have emerged from them as well.  When I make significant changes in what I said in this session, I will mark that by presenting them in the green type of this note.  I am not, by the way, sure that I have always gotten her corrections to my attributions of she has said correct.  So there may be revisions after you first read this.  🙂 

In fact, if you have the patience, it might be best, as I said in the announcing email, to ignore the green sections in your initial reading.  Such a reading will give you a better sense of the “flow” of my lecture, as given.  Then you can consult Marla’s further, often correcting, indications in green.

When I began to work with this topic, almost everyone I mentioned it to responded positively to it and said that it seemed an interesting one.

I think that is because it is a topic that throws up questions for most of us; questions we think it might be interesting to explore.

But it also indicates this topic’s richness, and the likelihood that each of us would be likely to see questions and issues in it that could take us in a variety of directions.

So I want to acknowledge that this is a talk that could be done in many different ways, and if you undertook it, the way it would turn out for you would likely be different, maybe very different, from that way it has for me.

I make no claim that my “cut” into the question of what makes rugs and textiles “Easy to Weave” or “Hard to Weave,” is an optimum one.

It is merely where this question has led me as I explored it.

So with that “apologia” in place, let me begin.

Although he is entirely innocent of it, this program is likely Walter Denny’s fault.


During a “walk-through” of his TM exhibition “Anatolian Rugs in the Classical Tradition,” a few years ago, Denny spoke a sentence that has turned out to be seminal for this session.

He was talking about some Turkish rugs with geometric designs that he described as “easier to weave.”  “Wait a minute.” I thought, “What kind of design is that?”

In truth, I can no longer say precisely, but I think Denny was referring to


this derivative of a small-pattern “Holbein” rug, or


to one with “Memling” guls.

Denny’s indication got me thinking more generally about what might make rugs “easier to weave, or, alternatively, “harder to weave.”

Now I am not a weaver,


nor am I a member of one of the weaving cultures within which the rugs we tend to collect were woven.


So what I am attempting in this “rug morning” could be dangerous, if not a bit foolhardy.

Still, we are routinely advised that one of the soundest activities we can undertake as collectors


is the close examination of the pieces we own, and this topic seems at least allied with effort of that sort.

And this set of questions, even if ethnocentric, and a bit ignorant, are among those I have.  So I have refused to apologize from my perhaps misdirected curiosity and have continued.  This program is the result.

It seemed to me that there are at least two vectors in terms of which levels of weaving difficulty might be arrayed and explored.

The first is the character of a given woven structure itself and how difficult it is to produce.

The second dimension of weaving difficulty is that highlighted in Walter Denny’s comment.  That is, what designs are easier or more difficult to weave.

I want, here, to treat both of these aspects of weaving difficulty in turn.

Let us first consider some woven structures, and try to estimate which are easier or more difficult to fashion.

We begin by ruling out some instances that seem to entail great complexity, but which are not an issue, mostly, for collectors of our day.

I refer, first, to complex fabrics that have two levels of warp.


Some fabrics made in 14th century Spain and during the 15th and 16th centuries in Ottoman Anatolia and Persia, are of this sort.

They are made on “draw” looms


that require simultaneous employment of two weavers.

An even earlier set of complex weaves are some of those employed by South American Indian weavers during pre-Columbian eras.


Here, “scaffolding” structures were sometimes used and items could be woven, resist-dyed, taken apart, and then reassembled in new combinations to produce complex designs and fabrics.  Some pre-Columbian textile structures are among the most complex we know.

I also knew that some very real difficulties of weaving setup, things that affect what Marla Mallett calls “weave balance,” are usually decided (often handled as well) by the local weaving community and tradition or by workshop managment.  So often the folks who actually do the weaving are held away from some very real aspects of weaving difficulty, comprehensively defined.

I began to think, just musing, intellectually, about what woven structures might be the easiest to produce.

The easiest woven structures to produce would, intuitively, seem to be plain-weave tapestries in which the wefts and warps are the same color.

drawingbalancedplainweaveWoven fabrics in which the warps and wefts are the same color do not jump at you when you begin to look, but I think there will be intances in most collections.  I have an Anatolian textile


that has large areas of flatweave with the same color warps and wefts in a balanced plain-weave


that is surprisingly open and lacy in close-up.


The ivory ground of this Anatolian textile is cotton and linen, a fabric,  an Anatolian dealer told me, very like that the Egyptians used to bind their mummies.

A second level of seeming difficulty is that in which the warps and wefts are different colors and different sizes and this difference in size will result in the structure being “weft-faced” or “warp-face.”

Let us take a weft-faced tapestry of a single color as our example of this level of difficulty.  Kilim ends of rugs are often of this sort.


These distinctions about what techniques might be easy or hard to weave are intellectual and my own.  They are distinctions that seemed to make sense to a non-weaver standing outside any weaving community.

But how would such distinctions seem to experienced weavers, much less to members of a weaving tradition in a rug-producing community?  The distinctions we have cited so far seem nearly trivially straightforward and might not be the occasion for much debate, but that could change before we go much further.  For example, do actual weavers experience and recognize degrees of difficulty in weaving particular techniques such as sumak, brocade and zili?

I needed to test with some experienced weavers my sense that there were likely levels of difficulty in weaving sourced in the different structures themselves.  So I contacted the late Peter Collingwood in England (in the weeks before his recent, unexpected death)


and Marla Mallett, in Atlanta,


and asked my “which weaving techniques are more difficult to weave?” question.

I fully expected that they would be able quickly to tell me whether “brocade” was more difficult to weave than, say, “sumak” or “zili.”  In my “man-on-the-street” innocence, I also expected that experienced weavers would, for example, say that techniques that can be woven from both the front or the back are easier to weave from the front (because you seem likely to be able to see what you’re doing more readily).

The answers I got surprised me.  We never got to the listing of technique difficulty because both Peter and Marla were cautiously critical of my basic question.  The question of whether a given textile structure is easier or harder to weave (because of the character of the structure itself) was one that did not leap at them.

Collingwood said:

“Do you mean to put them into a sort of table going from the easiest to the hardest?”

“Actually, I am very hesitant to advance definite opinions in the area…An expert in process A, who does nothing but that all his/her life, will have learned all the tricks, the shortcuts, and so finds it easier to weave that than process B, which to us westerners, seem easier, simpler, quicker.”

Marla’s comment was harsher.  She began by saying: “I can’t see that this is any distinction worth discussing and frankly it seems a bit silly.”

(That may be true, but to repeat, I feel unapologetic: one can only ask the questions that one has.  The most useful response is to explain why this is a bad question.)

Marla did provide some explanation of this latter sort.  She said:  “The weaves used by Western and Central Asian tribal weavers are all simple; none are particularly difficult once one has learned the processes and become aware of the inherent limitations.  They only become “difficult” when used in unnatural or unsatisfactory ways.”

Now there are senses in which Peter’s and Marla’s objections to my question of difficulty, asked “in the air,” so to speak, seem unimpeachable.  It is clear that most answers to this question seem to require answers to prior ones.  It depends, importantly, on who you ask this question of and how that person is situated.

As Collingwood points out, a 12-year old girl in a weaving community is situated differently from an amateur weaver in the U.S. who has his or her first frame loom and is just beginning.  A Shahsavan weaver who has been working with sumak for five years is situated very differently than is a boy who has been working for the same time in a Kerman “factory” tying pile knots in response to a caller.

So I want to acknowledge that proper answers to my question may well be highly contextualized.  But I think not entirely.

More, I may be mistaken, but it seems to me that there’s something wrong with this contextual argument.


Learning to do something well does not, I think, reduce its inherent difficulty.  One must still deal anew with that in each new performance.  What has changed is not the difficulty, but rather one’s skill in performing as desired in spite of it.  The difficulty is still there to challenge one’s skill forever in future performances.  Increased skill does not reduce difficulty, it merely equips one to overcome it predictably.

What is deceptive is that the exercise of increased skill may be experienced as reduced difficulty.  So, it seems to me, there is, at least, a debate here about what is the most appropriate focus for the word “difficult.”

(I apologize for what might be seen to be unneeded close distinctions in my counter-argument here.  They are the residue of my “hanging out,” once, for a little too long with some philosophers.  🙂 )

Both Collingwood and Mallett did make some further indications that encourage me to think that they, too, sometimes see some aspects of weaving difficulty that are not entirely rooted in context.

Before he wrote the paragraphs I have cited above, Collingwood was willing to venture this much with regard to whether it is easier or more difficult to weave from the front or the back of a piece being worked on, using a technique that can be woven from either side.  Collingwood said:

“I think it’s always easier to pass a hand-manipulated weft OVER two or more ends (ed. “warps”) than UNDER them.  So sumak and brocade are easier from the front…and what I call “skip plain weave” is easier from the back.”

Note: Collingwood’s “skip plain weave” (defined in his book,  Techniques of Rug Weaving) is one that has areas where the wefts pass over and under alterative warps, but also other areas where a given weft passes UNDER several warps before resuming the alternative warp sequence.  What he is saying is that this area, where the weft is passing UNDER several warps, is easier done from the back because this has the effect of converting the weaver’s task to passing these wefts OVER these several warps, something Collingwood’s rule suggests is easier.

Peter’s indications here are of the sort that I thought it might be possible to make about intrinsic difficulties of particular weaves.

In subsequent conversation, Marla Mallett also said a little more.  “…When you mention sumak and brocade, well brocade is done almost always from the back side.  Sumak can be done either from the front or the back—there are advantages and disadvantages to each, although the majority of extant examples have been done from the back…”

(This last point is interesting, since Collingwood’s impression is that both sumak and brocade should be easier to weave from the front than from the back.  Why do most sumak weavers and nearly all brocade weavers chose what seemed to Peter, the more difficult side from which to work?  There must be some other aspects of the “advantages versus disadvantages” equation, that Marla mentions, at work here.)

Note:  Marla said in subsequent conversation that Peter’s indication that brocade should be easier woven from the front is true if one is weaving with a single color, but that a real difficulty emerges if one attempts to brocade from the front using more colors. 

Marla acknowledged that there are situations in which it could be easier to weave brocade from the front and others in which brocade must be woven from that side.

In areas of brocading where colors continue across the piece and the unused colors “float,” weaving from the front would seem advantageous.

And Turkmen tent bands of mixed technique must be brocaded from the front because pile is  also being employed in some areas, and pile  knots (the pile of which projects from the front of the piece) can only be tied from the front.  Still, Marla notes, the task of weaving brocade from the front on such pieces is made easier by the fact that all the weaving is being done only on alternate raised warps.  The brocading is being done on the surface, so to speak, of the entire fabric being woven.  But the use of brocade in such mixed technique weavings is a special, and infrequent situation.

Marla also suggested that there might be a question beneath my question.  “…I suppose the question you might be asking in actuality is how long the learning curve might be for the various techniques…in this case pile weaving is surely the simplest and quickest to master.  That requires about 15 minutes.  And  (ed. with sumak) it’s the simplest in which to execute most designs.  Virtually anything can be copied in pile (ed. or sumak) while designing in each of the other structures is limited.  Knotted pile is also the structure easiest to find people to repair.  Most restoration people are terrible when it comes to flatweaves.”

Marla may be right that the question I want to ask is best couched in terms of learning curve length, but the instructional designer in me rebels at this.  “Length of learning curve” does not seem like a very plausible independent variable.  The reason it takes longer to learn to do something depends on other things.  The intrinsic difficulty of the required recognitions, selections, holdings and motions involved in weaving the various techniques seem more plausible as “root” reasons indicating why it takes longer, on average, to learn to weave one technique versus another.

But before I move to the second question of what designs are easier or more difficult to weave, let me offer:

1) an analogous argument from the arena of macrame, one in which I once spend some time as an interested practitioner, and

2) an example structure that Marla, herself, cites as one that may be an instance of a woven structure that is more difficult to use.

First, the analogous argument from macrame.

The two most widely used knots in macrame are the “square knot” and the “double half hitch.”  As Marla says about the most frequent weaves, neither of these two knots is difficult to “tie.”  But I want to compare them a bit here because I think one is intrinsically more difficult to “use” than the other.

Let’s take the square knot first.

Look at the sequence in the slide below.


It shows how a square knot it tied.  In macrame usages of the square knot, such knots are often tied around two “core” cords for a variety of reasons.  Ignore the core cords, marked “B” and “C” in these drawings,  and focus only on the outside cords, “A” and “D.”

The bottom slide in this sequence is what a completed square knot looks like before it is tightened.

Now  the square knot is not only simple to tie, it is, usually (within macrame usages) easy to use and to control as it is being tied.  It is true that, sometimes, when you are using a square knot to tie a package, that one has to ask someone else to “put their finger on” the first half knot to keep it tight and in place while one ties and tightens the second half.

But that problem does not occur in most macrame usages of the square knot and one result of that is that there is NO “control” problem associated with getting the second half of a square knot “in” and firmly tightened.  One needs only to apply a constant tension in tying so that the square knots you are producing are of the same size.

Now let’s look at the “double half-hitch.”  Here is a drawing showing how this knot is tied and its appearance when tied.


The double half-hitch is tied around a core cord.  Basically, two passes are made around this cord and the knot is only stable in place when the second one is in and tight.


Above is an image of  several horizontal rows of double half-hitches tied one under another.  Look at the right side of the bottom row.  Notice that the double half-hitches on the right side seem to be wandering away from a tight fit between rows.

Like the square knot, the double half-hitch is not very difficult to tie, but because the position of the knot is determined by the position of the core cord when the knot is tightened,

the double half-hitch is very difficult to control.

The is a great tendency for the position of the core cord, and hence of the knot, to wander while one is attempting to get the second loop in and tight.

The fact that the placement of a given double half-hitch is determined by that of the core cord when the second hitch is made tight, makes me rate it as considerably more difficult to use than the square knot.

Here is another example of the use of the double half-hitch that demonstrates the “fit between rows” difficulty and one other.


This simple checkerboard design is a small wallhanging that I tied in the 70s.  It is almost entirely of double half-hitch.

Look first at the upper left hand orange square.


This square is produced by tying rows of orange double half-hitches around dark brown core cords.  The first demonstration of skill with this knot is that each of the rows fits tightly against the one directly above it.  There is no wandering off between rows.

Now look at the second square from the left in the top row.

Notice that it is brown.  This is because the former brown “core” cords are being used as the tying cords and the orange cords now function as the core cords.  One result of this change is that the double half-hitches in the brown rows are turned 90 degrees from the orientation of the orange knots  in the first square.

This shift of knot orientation, in order to change the color of the second square, creates an additional difficulty of control.  One must now, not only insure that each double half-hitch being tied fits snugly against and directly below the one above it, one must also “fit” the reoriented brown square into the space defined by the shape and size of the orange knots.  This can be difficult if the shape of the orange double half-hitches is not entirely square.

There are 35 changes of color from one square to the adjoining one in this little piece.  Seventeen of these squares are brown.  So, this little piece, with its simple checkerboard design, is a demonstration of one’s ability to control the placement of the double half-hitch, not just by keeping the knots tight between rows, but by being able to tie every brown knot so that it fits into a space slightly different from that of its actual dimensions.


The fact that the placement of a double half-hitch is difficult to control also shows, I think, why much of the contextual argument about estimating the difficulty associated with weaving a given structure is faulty.

The difficulty of controlling placement of the core cord is always present in each new double half-hitch tied.  The difficulty itself is not at all diminished by increases in one’s ability to deal with it successfully.  Yes, “skill” has increased, but the difficulty is not thereby reduced and must be dealt with anew each time a new double half-hitch is tied.  For me, this argument, despite its analogous character, is telling about the liklihood of similar difficulties inherent in using particular woven structures.

Note:  Marla said that most woven structures do not entail, the kind of inconsistency, I described in my analysis of the effect of the orientation of the double half hitch.  But in another part of our conversation Marla seemed to suggest that sumak  may sometimes entail a difficulty that is somewhat like the inconsistentency associated with tying a double half hitch in two different orientations.  She said that like pile, sumak is digital and any design can be drawn using it.  But if the design being woven in sumak requires the wrappings to “return” at the edge of a design device the edge of the design device will be slightly ragged looking.  Marla contrasted this characteristic of sumak with that of “reciprocal brocading” which is an “interlaced” structure rather than a wrapping.  She said that some Anatolian applications of “reciprocal brocading” are very uniform and do not exhibit the inconsistency that sumak can at the edge of devices.  (I have, additionally, wondered whether “plain sumak” in which the weaver has to employ specific measures to insure that the front-face wrappings slant in the same direction, is not slightly, but intrinsically, more difficult than “countered sumak,” which is what is produced if one simply continues wrapping one row after another.  Marla and I did not talk about this latter possibility.)

We also did not discuss whether Marla sees one other  instance she identified as of this sort, but it seemed suggestive to me.  Marla said that it is more difficult to use sumak to weave small devices than is the case with pile weaving.  There is no difficulty with  tying a single pile knot  on a pair of warps (“Spanish” knots can be tied on single warps) but it seems more difficult to use wrappings to weave small devices.  Wrappings are digital, but not as independent of one another as are pile knots.  This connectedness, it seems, gets in the way of using sumak in small devices.

Marla did not say why sumak is hard to use for small design devices, but it seems to me that the fact that sumak is a series of wrappings ,means that the integrity of each wrapping is dependent, in a way a single pile knot is not, on its connectedness with wrappings on either side  of it.  This need for continuation of the wrapping weft is a disadvantage when small designs devices need to be drawn.  The problem is how to hold a single wrapping of a single color in place.  Not that it can’t be done, but that special provision needs to be made for it.

Are these miniscule but real difficulties intrinsic to some uses of sumac?

Now let’s examine Marla’s instance of possible intrinsic weaving difficulty.  The reason what I felt that weaving from the back might be expected to be more difficult than weaving from the front, was that I assumed that the weaver usually had more difficulty “seeing” the results of his/her work, and the additional difficulty of determining what to do next with which cord of what color.

Marla seems to acknowledge that she sometimes has a similar concern with weaves that seem to block the weaver’s view almost completely as he/she work.

She said in this context:

“…To my way of thinking, weft substitution, which is also done from the back, is the most ‘mysterious’…I am always a bit amazed that those Moroccan Berber women know what they are doing when they can’t see the pattern at all!  (Just take a look at the back of a Baluch weft substitution piece.)  But the process simply becomes automatic, with the position of each new weft just based on the interlacement pattern of the previous one.  My understanding of it surely results from the fact that it is a process that I’ve never used myself, beyond simple experiments to make sure I understood it…”

Here is a detail of the front of a Baluch piece that exhibits weft substitution:


And here is the back of this same detail.


This morass of floating wefts seems difficult to discern and the ability of a weaver working on this side to select that right cord and color and to know what needs to be done next with it, does seem to merit Marla’s word “mysterious.”

Notice that Marla has described the cues these weavers follow, from the back.  To quote her again, “…the position of each new weft (and this is my addition: its color) is based on the interlacement pattern of the previous one…”

I think with the instance of weft substitution Marla is acknowledging that there are aspects of the question of the intrinsic difficulties entailed in fashioning particular woven structures that have attracted her attention as well.

There are cues about what the weaver should do next with what thread of what color and they reside in the “interlacement pattern” of the previous weaving action.  They are apparently subtle enough that a weaver of Marla’s experience cannot (without working with the structure for a time) discern them.

My own view is that this suggests that weft substitution weaves are more difficult to fashion than are some others.

Note:  In our subsequent discussion Marla suggested that my characterization, above. of weft substiution needs to be modified.

First, the image of a detail of completed weft substitution in the Baluch piece above, is not an example  this structure woven from the back, as the Moroccan example she cited was.   The clue is the narrow bands of border that resemble weft twining.  Marla calls these narrow bands instances of “wrapped and bound” borders.  These narrow borders can only be woven from the front of a piece.

Weft substition weave is done with “complementary” wefts of different colors.  When the  first weft of such a pair is woven its role and consequence must be calculated across the entire row for the design intended.  This can be difficult.  Conversely, the second of such a pair of complementary wefts is easier to weave since it moves over and under warps opposite the alterations of those of the first weft.  Its job is follow an interlacement pattern opposite that of the first weft to “fill in” the design pattern established by the first weft.

This explanation suggests that generalizations about difficulties, of flatwoven structures in particular, are difficult.  Some aspects of a given flatwoven structure can be pretty straightforward, but other aspects of weaving the same structure can be difficult. 

While acknowledging that generalization is difficult, and perhaps sometimes not possible,  weft substitution does, on balance, still seems to entail subtleties of recognition that, to me, at least (and I have never attempted even the experimental efforts to use this weave that Marla has) suggest that this is a woven structure more difficult to use than some others.

In our most recent conversation, Marla seemed to make some indications that might license other judgments of some weaves being more difficult than others.  Sometimes it seems the difficulties associated with using a particular structure have implications for drawing a particular design. 

Weft substitution seems to be less flexible for use in design than is, say, brocade, a technique with its own restrictions.

I think if we talked longer and experimented a little more with how particular structures are woven we would find ourselves describing particular situations in which there are intrinsic difficulties.

To conclude on this first dimension of possible weaving difficulty:

o  Peter Collingwood’s “over is easier than under” comment;

o  My macrame analogy;

o  Marla’s weft substitution example; and

o  Possible defects in the “contextual” argument,

combine to make me wonder whether my suspicion about possible intrinsic differences in difficulty among particular woven structures does not have some merit.

Still, I would not have predicted that this context aspect might be, potentially, so important when I began my work.

Note: Marla’s suggestions about some faults in my description and argument above suggest to me that some erroneous aspects need to be  corrected, but I still think that my suspicion of intrinsic difficulties associated with particular woven structures may be sustained.


Let’s move to examine aspects of weaving difficulty that are the result of design differences.

I have drawn for some of my points and distinctions on two typologies and some email exchanges with someone who has begun to weave in recent years and who has to date woven six pile rugs.

The first is Claude Humbert’s source book on Ornamental Design, 1970.

Carol Bier’s analysis of symmetry and pattern, initially introduced in a Textile Museum exhibition a few years ago, provided a second source of relevant analytical insight.

The U.S. weaver on whose experience I have drawn is Mark Traxler.


He is by himself in Minnesota and has built his own loom, done some of his own spinning and natural dyeing, has composed cartoons of designs he wants to weave and has then woven them.

In this treatment of what designs might be more difficult to weave I want to exclude approaches to weaving that finesse the problems associated with weaving (largely) from memory.

First, I want to exclude pile weaving the result of someone calling out to the weaver(s) the color of each knot to be woven.

Similarly, I want to exclude weaving that is performed by closely following a written guide or cartoon of some sort.

I want to treat, centrally, weaving done by  weavers who have to rely primarily on their memory of the design they are weaving.

Note:  In our subsequent conversation Marla added that the weaver who is doing real design as he/she goes along, is doing something very different from the weaver who is strictly following a memorized design.  The weaver who is being “creative” at all must deal ,not just with drawing the intended design, but also with the relationship between the changes he/she is making and restrictions of the particular weaving technique being use.  This latter is, potentially, a far more formidable undertaking.

Now, in truth, it is difficult to detect accurately the point at which the complexity of design requires the move to a caller or to a cartoon.  So to some extent we are guessing about which of some more complex designs might have been woven from memory.

For awhile, on my observation of a DOBAG weaver in Turkey, I thought that weavers must, like champion chess players, have a knot-by-knot mental image of a given design on the bare warps.  What made me think this was that this DOBAG weavers put in all the knots of one color in the row being worked on, before moving to any others and did it with great speed.  She moved to pairs of warps, at some distance from one another, quickly and assuredly.

But Marla Mallett and Mark Traxler disabused me of my impression.

Marla said in this context:

“It is necessary for any weaver to figure out how many pairs of warps each motif requires, and to position the first knots properly.  Then each succeeding row of knots is tied with their positions based on the row before…either vertically above, or one pair to the right, to the left, etc.  Or three pairs to the right, or left, etc.  The position of each part of each row is always based on the position of the knots in the row before.  Then when beginning new motifs, the starting point of each position has to be calculated.  I think if you tried it yourself, you’d understand completely.  I don’t think any mystical “digital image” is involved…just a series of calculations, with each based on the parts just finished.”

Both Marla and Mark indicated that even quite complex designs can be woven, eventually, and I will show some surprisingly complex examples.

Note: In conversation, after my TM presentation of this lecture and session, Jim Henderson,


(that’s a not very revealing image of Jim on the left in the photo above) told me of an experienced weaver who has shown how weavers “break” down complex designs into parts, sub-parts and sub-sub parts that permit them always to be working with quite simple elements easy to hold in memory as one works.  Jim said he has not been successful in getting this weaver to write down what he knows, but we plan to make another effort, perhaps, by interview.

I think you will be able to argue, sometimes, when we get to my more complex design examples, that I have not always been able to abide by my “woven from memory” standard.”

Note:  Marla indicated in our subsequent conversation that I, in fact, HAVE included a number of “cartoon” rugs among my examples.  She pointed especially to some Persian and Chinese examples below.

But that is my intent.  I have, for example, excluded pieces with resolved corners, since that feature seems conclusive evidence that either a caller was used or that a cartoon was followed closely.  I have taken “butted” borders to suggest that a piece with them COULD have been woven from memory.

In what follows, I will quickly list and exemplify some indicators that seem to me to make it likely that particular designs are easier or more difficult to weave.

To some extent, the result turns out, largely, to be close to what one might expect even without giving the matter serious thought.  Still, an explicit listing of even the seemingly obvious might be useful, since it brings particular indicators to self-conscious light and makes them visible for critique.

Normally, in these “rug mornings” we are very interested in such things as attribution, age, materials and dyes, but if I did that I would have to break at this point and serve some wine and cheese and then take another hour to finish this lecture before we even got to looking at the pieces in the room.

So you need to shift gears a bit and put on your “TV-beer-commercial” eyes, since I am going to mention what I think are fairly obvious indicators of design weaving ease and difficulty and literally flash past you, without much comment at all, examples of each.

Some may not be satisfied with that and, if you insist, I am glad to make an appointment to take you through another version of this presentation I have that includes such things.  But you will have to buy the wine.

Let’s begin with a couple of indicators of designs that seem easier to weave with a couple indicators that are fairly pure categories.


Here are some examples:





It is probably not useful to complicate one of the clearest indicators of designs that are easy to weave with an exception, but there is one here.

There are some Moroccan textiles that are woven partly in wool and partly in cotton and entirely in white.

These pieces are dowry pieces, dyed only after marriage,


and it is the differential absorption of the dyes by cotton versus wool that reveals, ultimately, the designs that are there from the beginning, as the result of  the systematic use of wool in some areas and of cotton in others.

The feel of the wool versus that of the cotton is readily recognizable, but the weaving of them “white on white” to produce fine-grained designs without the aid of color differences,


seems to me not to be an easy undertaking.

When we move beyond areas and blocks of a single color, we encounter a slightly different and more difficult design feature: bands of varying color.


At this level there is no real “drawing.”  The design decisions to be made are:

1)  What colors are to be used?

2)  Which colors are to be used next to one another? and

3)  How wide or long should each stripe be?

Because this simple design alternative is available to nearly all weavers, it is widely used.  Here are some examples:



The khorjin set above is warp-faced, Shahsavan of silk and wool.




The striped Navajo blanket above is in The Textile Museum collection and was shown, recently, as part of the “Blue” exhibition.  This image does not nearly do its colors justice.

The fact that what is desirable to contemporary collectors is often not correlated closely with designs that are difficult to weave is signaled by the fact that a similar “first-phase ‘chief’s’ blanket” sold at auction in recent years for over a half million dollars.

Notice, though, that quite sophisticated aesthetic effects can be achieved even at this easy level of design weaving.  This is a place where Humbert’s notion of variations in “rhythm” in the use of design devices, even as simple as the stripe, draws attention to an aspect of design we have surely seen, but not noticed adequetely.

Here are two views of a graphic Humbert uses to illustrate his notion of “rhythm” in ornamental design.



You can see that the frequency, length, width, , orientation, placement and colors of stripe designs can affect their complexity in remarkable ways.

I own the back of an Anatolian grain bag that I bought before I knew what it was


(this photo does not show off its actual colors)

and find that I never tire of looking at its stripes of color.

For me, the weaver has managed something not reflected at all in the easy weaving task of creating horizontal stripes.

Some aspects of textile design that can be used to distinguish those that are harder from those that are less so, are actually ends of continua.  Simple versus complex, rectilinear versus curvilinear, etc.  The next few indicators I will treat are of this sort.  Again, I am going to tick through some examples without describing them much.


Here are some simpler designs.





The piece above was made, largely, of corn husks by U.S. “plains” Indians.


Below are examples of more complex designs.



A closer detail of the piece above.





Here is a closer detail of  a corner of the “nomad Kerman” rug above.


The next continuum is “rectilinear” versus “curvilinear.”


The grid of warps and wefts on which textiles are woven is rectilinear.  So designs that comply with the character of this grid should be easier to weave.

This might be the place to indicate and demonstrate just how powerful this tendency to the rectilinear is.

In his study of tribal rug designs, Peter Stone provides the graphic sequence below demonstrating the effect of the reduction of the number of knots per square inch can have on a circle produced at a higher knot count.


As the labeling indicates, on the left is a circle woven at 1600 knot per square inch.  The next figure to the right is this same device woven at 400 kpsi.  The third example from the left is what the circle becomes when woven at 100 kpsi.  Last, the circle becomes very rectilinear when woven at 25 kpsi.

We shall see that weavers can sometimes create credible, visually circulinear designs at surprisingly low knot counts, but the above sequence demonstrates how strong this tendency to the rectilinear is.

Here are some examples of designs that are rectilinear and, therefore, easier to weave.






And here are some examples of the curvilinear end of this continuum.  You may disagree that some of the more complex designs below were woven from memory, but let’s see.  These designs should be more difficult to weave.


The achievment of a curvilinear effect in the 17th century piece above is noteworthy since the knots are large and the kpsi is relatively low.



A closer detail of the piece above.




The next continuum of design weaving difficulty I want to exemplify is designs composed of SIMILAR devices versus those that feature DIFFERENTLY SHAPED devices.

DESIGNS COMPOSED OF SIMILAR DEVICES WOULD SEEM, GENERALLY, EASIER TO WEAVE.  (It helps if there are only a few and they are larger.)

DESIGNS COMPOSED OF DIFFERENTLY-SHAPED SHOULD BE MORE DIFFICULT TO WEAVE.  (Difficulty is increased if there are lots of them and they are more detailed and smaller.)

Here are some examples of designs with similar devices.






Here are four examples of designs with differently shaped devices.


A detail of the piece above lets you see the number and variety of devices better.



The curvilinear medallion above reveals the detailed differences that even a single device can exhibit.



The next continuum is designs that employ a narrow color palette versus those that use wider ones.



Note:  Marla disagreed that designs that require wider color palettes are necessarily harder to weave.  She pointed out that the larger number of colors often provide more markers indicating, usefully, where the edges of devices (and their internal instrumentation) occur.  On the other hand, she cautioned that this effect of the use of more colors is a rather mild one and that the seeming converse: that designs with fewer colors might be harder to weave, is, in her view, not the case.

Here are some designs that require few colors.




A closer detail of the older Chodor rug fragment above.


And here are designs that require wider color palettes and should, therefore, be more difficult to weave.

Note: Again, notice Marla’s disagreement with this rule.



The above Hamadan Tafresh has 15 colors.


Despite the fact that it was very likely not woven from memory


I want to include the detail above from Wendel Swan’s wonderful Kerman meditation carpet, (you’ll see it all in another category below) to illustrate how range of color, for which Kermans are famous, is exhibited in it, and works to make its already difficult design even more difficult to weave.

Note: In our subsequent conversation Marla pointed out that since Wendel’s rug above is very likely a cartoon-based piece, the issue of the wider color palette and its impact on difficulty of weaving this design simply doesn’t apply.  The weaver is merely following a cartoon closely.

One last thought about color.  Edwards, in his famous book on the Persian carpet, admires the work of Heriz weavers.  He says that they often work from simple, handkerchiefs, printed with curvilinear designs in two colors.  For this image, he says, they weave rectilinear versions in a dozen colors.

Now a correctly-spaced conversion of a curvilinear design into the rectilinear is not a feat to be under-rated, but Heriz rugs tend to have low knot counts, and, given what Stone has shown about how strong the pull of the basic grid of warp and wefts is toward the rectilinear,


the real difficulty might be rendering a two-color handkerchief in twelve colors.

herizoverallcolorsdetail The next continuum of design weaving ease versus difficulty moves between designs that are spacious and those in which the devices are densely arranged.


Note: Marla disagreed with this generalization for reasons similar to those in her observations about color palette.  In fact, she points out, dense design devices provide more points of reference for weaving.

She also observed that it depends on what structure is being used.  With warp-faced structures like that of most jajims, the colors are in the warp set up and, if the design is composed of small repeats, quite dense designs can be easy to weave.

With brocading, she said, a “compact structure by its nature,” denser designs are actually easier to weave.

Marla indicated that the frequent mistakes one can find in design placement in older Turkmen is likely the result of the valued spacious layout.  (If this is the case, then one should find fewer placement errors in Turkmen pieces when they became finer and more densely patterned at the turn of the 20th century.)  

Here are some spacious examples.


A closer detail.




A last spacious example is this detail of a 16th century Ushak saf.  There are six such panels in this piece.


Designs arranged densely should, generally, be harder to weave.

Note: Again, remember Marla’s substantial disagreement with this this rule.




Carol Bier drew attention to the notions of rotation and reflection in her Symmetry and Pattern exhibition at The Textile Museum a few years ago.

A design is “rotated” when it is moved 90 degrees while a point at one corner is held in place.  A design is “reflected” if it is moved 180 degrees along either a horizontal or vertical axis.


Here, below, is a complete Hamadan rug woven in the Mehraban area that was probably not woven from memory


but also could, plausibly, have been woven in that way.

It is a design that can be produced in its entirety by either reflection or rotation of one of its halves or quarters.  This is the case because all of the design elements in each of these quarters is the same.

Here is a demonstration of how this piece can be produced by, first, reflecting one of its quarters on the horizontal and then by reflecting the resulting lower half vertically.

Let’s say that we have in memory all of the design elements of the lower left hand corner of this piece.  Our memory of it will look like this:


So we begin weaving the first row on the very bottom, moving from the left until we reach the center.  To go on with this design, we merely look left at the knots we have tied so far in this first row and tie them again, this time in reverse order.

This produces a right lower corner that is the mirror image of the lower left corner…hence the notion of “reflection.”  If we continue line by line we will eventually have produced both of these lower quarters and our rug will look like this.

reflectionrotationexample2ndquarteraddedWe continue now by tying the knots in the upper half of this rug, knot for knot as we have in the first half, but now we take our guide from the ROWS BELOW, working in vertical reverse order.

If we do this accurately, we will eventually produce the entire design which again looks like this.


Some designs can be woven entirely only by reflecting a half.  This is because the designs in two quarters are different from the designs in the other two.  “Niche” designs are often of this sort.  So are designs with lower elems.

Here are some designs that require rotation or reflection of a vertical or horizontal half.



The Aleppo horse cover above is woven in silk and metal.


Drawing in a rotated or reflected position is not always easy to bring off.  Many niche designs are woven upside down, beginning with the niche end


since it is easier to make adjustments at the other end, if that is required.

This, in turn, means that if there are design devices like birds or animals or ewers in the field, the MUST be drawn upside down


so that when the completed piece is displayed with the niche at the top,


So while the possibility of drawing in reflection or rotation makes a design easier to weave than one where this possibility does not exist, drawing in different orientations can itself be a source of increased difficulty.

But, to repeat, that does not alter the basic fact that, a design that CAN be woven entirely by reflection or rotation of a quarter or half is easier to weave than one that cannot.

Here are some designs that cannot be produced by reflection or rotation of a half or quarter.

harderallquartersdifferentkirman-tree-of-life-4This is a complete view of Wendel Swan’s Kerman meditation rug that I showed a detail of earlier.  I do not claim that this piece was not woven without close reference to a cartoon.  I include it because it is such a good example of a design in which all four quarters are different in some respect.


The contemporary rug above is 20 feet long.





I place rugs with designs that cannot be produced completely using rotation or reflection of a quarter of half among those that are more difficult to weave.

Note: Marla pointed out that although the rule about rotation and reflection may be accurate, it is useful to note that most moves of this sort are instances of reflection and that resort to rotation is relatively rare. Rotation seems to be employed mostly in border designs. Interestingly, differences between upper and lower borders, on one hand, and the same designs used on the sides that look different are often not the result of failure of drawing in a rotated fashion, but rather the result of the fact that the vertical-horizontal knot ration is not close to 1 to 1. If the vertical to horizontal knot ratio moves away from 1 to 1, even a very accurately rotated border design will look very different on the sides than it does on the top and bottom of the weaving.


As I worked with these two dimensions of potential weaving difficulty in my preparations, I found that I had to revise my beginning assumptions…repeatedly.

The strong influence of context on estimates of intrinsic difficulty in fashioning a particular weaving technique seemed nearly disabling.  I still think there is something to be said here, but it likely needs to be said by a weaver,


preferably one working within a traditional weaving community.


I leave to you the judgement of whether I have made a beginning argument for this possibility that is at all convincing.

In the case of what designs seem more difficult to weave, I was surprised, perhaps more than I should have been, about how commonsensical and obvious the key variables seem to be.

Note:  I think I need to qualify the conclusion above.  It seems to me that Marla has pointed out some instances in which “common sense” does not have sufficient access to the weaver’s experience and so can be erroneous.

Carol Bier’s analysis draws useful attention to the fact that designs that can be produced on the basis of reflection and rotation of one half or one quarter


are likely easier to weave than those that cannot.  But we also saw that the need for reflected or rotated design devices can itself be the occasion for increased difficulty.


And Humbert’s notion of “rhythm” in design helps us self-consciously identify and describe an aspect of design that we have certainly seen, but have often not noticed adequately.  It was slightly surprising to find ourselves estimating that designs that employ devices in irregular or unexpected rhythms do not, necessarily, seem noticably harder to weave.


It is useful, I think, to remind ourselves that the most sophisticated designs could be, and often were, produced by weavers, often children, who were largely interchangeable units of labor.  The “difficulties” associated with the most sophisticated designs have usually been dealt with by the designer or by workshop management long before the weavers come on the scene.

But it comes as no surprise that, for a weaver, weaving from memory, designs that are more complex, more curvilinear, with more differently shaped devices, that are more densely arranged, a that require a wider color palette, and that cannot be entirely produced by rotation or reflection of a quarter or half, are more difficult to weave.

Note:  I say the above, acknowledging that parts of it require revision based on some cogent points of disagreement from Marla.


I, also, think that it may be useful to note that the weaving required to produce the textiles that most of us collect nowadays is often of the easier sort.


especially since most of the decisons and skills required to establish the desired “weave balance” have been undertaken and determined by the local weaving tradition within which the actual weavers are working.

Even particular tasks, such as building or warping the loom are often held away from most weavers and assigned, in various divisions of labor, to more experienced members of the local weaving community.


Usually, when we praise the weaving of a piece we have collected, I think we are pointing, more than we self-consciously acknowledge, at aspects that did not require high levels of skill on the part of its actual weaver.

This is not to say that there are not gradations of weaving skill.  There assuredly are,


and in any weaving community the best weavers are readily identified by their peers.

It is just that the distinctions actual weavers make and notice are likely often more subtle than, and different from, those that are more central to our collector descriptions and evaluations.

This is the end of my lecture.  We moved now to illistrative pieces that I had brought in as well as those brought by members of the audience.

To see that part of this “rug morning” you need to go to Part 2 at this link:


R. John Howe

“Easy to Weave; Hard to Weave,” Part 2: The Pieces Brought In

Posted in Uncategorized on May 19, 2009 by rjohn

Dear folks –

This is the second part of a two-part virtual version of a Rug and Textile Appreciation Morning (RTAM) that John Howe gave at The Textile Museum on February 28, 2009.

If you want to start with Part 1, which is the lecture, you need to go to this link:

But if you have already been through the lecture, this second part presents and comments on both some of the pieces John


had brought in to illustrate the categories provided in his lecture, and those brought in by members of the audience. 

Since this is a topic that triggers different people in different ways, some of the pieces brought in by those in the audience were examples of different aspects of this question from those John treated in his lecture.

John began with two macrame pieces he had used to make his analogous argument, suggesting that some woven structures seemed intrinsically easier or more difficult to fashion.

The first of these was a small, compartmented hanging with a sampler design.


John used the center panel in the top row of this sampler to show what a tightened square knot looks like.


The sides of the square above are chains of square knots tied one on top of another, but the “wall” of fabric within the square is the best place to see tightened square knots.  Individual square knots occur at each junction.

He also had the checkerboard piece in the room that he used to make his argument about how difficult it is to control the double halfhitch.


 He passed both of these pieces through the audience.




Early in the “in-the-room” discussion, Pat Riley,


who has traveled widely, and observed some contemporary, but traditional weaving communities, commented that many of the difficulties of weaving seemed, in her experience to be associated with various aspects of setup and with techniques such as those that assured that warps were depressed to various degrees or not.

John agreed


and ticked off a list of factors that affect weaving difficulty, importantly, that he had not treated explicitly in his lecture, for reasons of time.

He said these include:

1.  What materials will you use to create your woven textile?  Wool, cotton, linen, silk, etc.?

2.  What sizes will the strands be that are used for the warps, the wefts, and (if a pile weaving) the knots?  How many plies will each of these strands have?  Will the working strands be composed of more than one of these materials?  (Even if you are not spinning these strands you will have to select them.)

3.  How many warps per horizontal inch will you use?

4.  How many structural wefts will you use and will they be placed in separate sheds or not?

5.  Will the structure be one with the warps all at one level or will alternate warps be depressed to some extent?

6.  If alternate warps are to be depressed, how much depression is desired and how will that be accomplished (will you “crowd” warps or or use taut wefting to displace them, or both?)?

7.  If a taut weft is used, will additional sinuous one be also, and, if so, how will they be placed in the various sheds?

8.  If this is a pile weaving, what knot will be used for the pile?  Will you vary the knot or its employment in particular situations and, if so, when and how?

9.  What type loom will you use?

10.  How (mechanically) will the warps be place on the loom?  And if there is provision for tightening the warps, as one goes along, how is that adjustment to be made?

John said that he had employed Marla Mallett’s term “weave balance” to refer elliptically to such aspects of weaving difficulty. 

He quoted Marla, saying that “A perfect balance must exist between the yarn sizes of warp, weft and knot, as well as several (ed. other) construction features.” 

John said that these are admittedly aspects of weaving that entail real difficulties. 

He said that he had (ultimately) omitted them from explicit treatment in his lecture because, as Marla points out, weavers, working within an established weaving community, have such aspects of weaving “handed to them,” so to speak.  

It is potentially very costly to experiment with such aspects of weaving.  Marla:  “Rather than tamper with a good thing, village and nomadic weavers tend to replicate the distinction weave balance worked out by the people before them. ” 

Weavers tend not to “play” with the solutions to the problems of weave balance their particular weaving community has some to over the years. 

This is one of the reasons why structural factors can often be an aid to attribution.

The next pieces shown were some of those used to illustrate various parts of the lecture that treated what designs are easier or more difficult to weave.


In the image above, on the board, beginning on the left, was the Anatolian weaving with Manastir-like end decorations, the plain ivory areas of which were John’s example of a balanced plain-weave in which the warps and wefts are the same color.

Peeping out of the edge of this Anatolian weave is a Yomut chuval the plain elem of which was John’s example of a weaving that employs a single color.

The kilim to the right of this chuval, was John’s “in the room” example of woven stripes.  It is the back of an Anatolian grain bag that he bought on the basis of the appeal of its stripes alone, without knowing that it was only part of a more complex whole.

The piece on the right in the image above is a Swedish rolakan hanging with a large-scale cruciform design.  It was John’s example of a simpler design.

The next piece was an opulent chuval fragment with lots of silk and a mina khani design.


 This fragment was John’s “in the room” example of a more”complex” design.  (It would have until recently been called a “Beshiri” piece, but is among those likely better described now with the term “Middle Amu Dyra.”)

John’s example of a rectilinear design was this Anatolian bag face in a zili weave.


Its rectilinear design seems among those that are easier to weave.

Wendel Swan had also brought a  colorful zili example.


A closer detail.



Wendel said this this design seemed to him to be one of the easiest to weave.

Note:  Marla Mallett said in our subsequent conversation that although zili looks simple to weave, there must be something difficult about it, because if you examine almost any piece woven in it, you are likely to find lots of errors.  Zili, she noted, requires careful distribution of color.

John had brought two examples of curvilinear designs.

The first was the contemporary Tibetan horse cover below.


The Chinese mat below has a curvilinear device in its field.


The curvilinear elements in these two pieces, especially those in its border of the Tibetan horse cover, were doubly difficult to weave, since both were drawn using relatively low knot counts.

Colin England,


an actuary, attracted to the drawing of curvilinear designs on a rectilinear grid, and hence to finely woven, often silk rugs, spoke to four examples, illustrating degrees of fineness, and how materials affect that (often by 0ften limitation).

His first example was the Saruk carpet below.


This handsome piece was the least fine of his examples and was woven in wool on a cotton foundation.  A cotton foundation is somewhat bulkier and places some limits on fineness.

Colin’s second “materials-fineness” example was the carpet below.


This carpet is wool-on-wool permitting more fineness than the wool-on-cotton Saruk example.

A third “materials-fineness” example was the rug below.


This pictorial rug is woven with wool on a silk foundation, permitting a fineness greater than that of the previous two.

Colin’s ultimate “materials-fineness” example is the sizable Hereke rug below.


This rug is silk knotting on a silk foundation, something that permits a fineness of 2,000 knots per square inch and above.  Only pashmina and silk can achieve the fineness that silk on silk permits.

One of the difficulties that confront weavers weaving complex, curvilinear designs at this level of fineness must be what can be seen by the naked eye.  I am not sure that magnification is not sometimes used, but its seems unlikely that it was in such weavings as the early Mughuls, that sometimes have a fineness of 2,000 kpsi and above.

John’s next “in-the-room” example was of designs with devices that have either similar or disparate shapes. 

His “similar” example was the “Ersari” khorjin on the left in the image below.


This piece employs devices that are similarly shaped.  It is, in fact, surprising, as one continues to look at the piece, at the visural richness and complexity the weaver has managed despite this.

The weaving on the right in the image above was John’s “in-the-room” example of a design with differently shaped images.  Here it is below by itself.


This piece features several differently-shaped devices, placing it among pieces rated more difficult to weave.  On the other hand, these different devices occur (excepting for the borders) in horizontal rows, thus permitting the weaver to deal mostly with one design device at a time.  This latter feature reduces somewhat the difficulty of weaving this varied design.

By the way, his piece, is not a rare bag-face, but one chest tab from a horsecover. 

One feature that has puzzled folks who look at it is that , despite being part of a larger assemblage, it is not “cut,” but is, rather, complete as it came off the loom.  Most suggest that it is likely Kurdish, maybe Bijar.

We now moved to pieces with narrower and wider color palettes.


The piece on the left, in the photo above, is a contemporary “faux pile” Siirt with four undyed colors.  It was John’s example of a narrower palette.

The Kordi “pushti” on the right does not have an exceptionally wide color palette,


but contrasts well with the Siirt piece.

John next treated designs that were spaciously arranged versus those the devices of which were arrayed more densely.  He had two spacious examples.


The first of these was the yellow-ground Konya fragment above (yes, the grass indicates that this photo was not taken in-the-room”)

A second spacious example was  a tired,  but dated Kazak.


This piece has the same design as a larger one with wonderful color in Schurmann’s Caucasian book. 

John said that this piece is one that we would not ordinarily show in a “rug morning,” but that the drawing in both its field and in its archaic-seeming border provides  good examples of devices spaciously used. 


This piece is dated 1319, which places it in the early 20th century.

John’s example of a design with densely arranged devices was the composed Coptic piece below.


 This is an array recomposed from parts of Coptic garments.  The “bird” forms in the “borders” are open and spacious, but the field devices are packed-in, densely, and may include some humanoid forms.  The seller estimated it as 4-5 centuries A.D.

John had brought a yastik woven with a “little rug” design as his example of Carol Bier’s description of designs that are easier to weave since they can be woven entirely working with only a half or quarter.


The design in the yastik above is one of those that can be woven completely if one has only one quarter of it.

John’s “in-the-room” example of a design that can be woven by reflecting one half was the Ladik niche design that he employed in the lecture.


John noted that he referenced this rug to show that niche designs were often woven upside down and that that  required particular design devices like ewers to be drawn upside down so that they would be oriented right-side-up when the piece was shown with the niche at the top.

But, in fact, this particular rug was woven with the niche at the top and no upside down drawing of the ewers in it was required.


But this piece is a good example of a design that can be woven completely by reflecting one vertical half.

Pictorial rugs are often among those that cannot be woven by rotating or reflecting a quarter or half.

John had brought a Firdowz as his example of this type.


As you can see each of the quarters of this piece is different.  So despite the fact that it is fairly coarse, it is placed among designs that are more difficult to weave.

We now moved to other pieces that had been brought in.

Tom Xenakis spoke to a jajim.




Although this warp-faced piece was woven in narrow strips and then sewn together,  the designs in it seem dense and complex enough to support an estimate that it was harder to weave.

Here is a closer detail of its front and back.


Note: Marla said that although this jajim is warp-faced, it is not the sort of design: dense, composed of small repeats, that she rates as easier to weave.

This jajim design, she said is one of those that she would rate as more difficult to produce because it is composed of larger design elements that are not closely arrayed repeats. 

Harold Keshishian had  brought in a horse cover that he felt, tentatively, might , be Senneh.


Here is a closer look at this “over the saddle” piece.


 The density and variety of the design elements in this piece suggest that it has a design worthy of being placed among those that are harder to weave.  Saddle covers were special occasion weavings and are argued to often have been made by more experienced weavers and from superior materials.

There might be some question of whether this dense design could have been woven from memory, but I think I have picked out in the image below the basic repeat from which it is composed.


Notice that this possible repeat is composed of simpler elements and would not seem impossible to hold in one’s mind as one worked.

The next “brought in” piece was this one.


My notes do not indicate what aspect of weaving ease vs difficulty this piece was intended to illustrate, but it has a rectilinear designs, pretty intensely arrayed, in some areas, but also features some horizontal banding.  


The rectilinear character of its designs, the relatively few variations in them, and the banding, argue for its placement among flatweaves that were likely easier to weave.

The next piece was the sumak mafrash side panel below.


Wendel Swan spoke to this piece, saying, that for him, it was an example of less than successful drawing.

Here is a closer look at it.  Can you spot the drawing “defect” that Wendel sees?


 Wendel points out that the “core” from which the blue latchhooks move out on both sides is not smooth, but jogs frequently.  And the “opposing”  (inside and outside) blue latchhooks are sometimes not smoothly aligned with each other. 

I think Wendel sees this piece as one that could have been drawn better and that shows signs as having been difficult for its weaver.

The next piece was also brought in by its owner as an instance of poor drawing.  It was a complete khorjin set.


Its owner described the drawing of the two medallions on it as “wonky,” apparently thinking that they were too different from one another.  Most in the room, including Wendel, disagreed.

This piece had good color in its pile faces and an unusually long, and especially colorful, connecting panel.


Wendel Swan had also brought in a Persian tea purse.  These items, still being used, are rarely seen by U.S. collectors.


It is placed among those that are more difficult to weave, both because of all of its quarters are different, the design is curvilinear, and because its small size likely makes drawing more difficult.

He said that this one is 9.25 inches V and 7.75 inches H.  It has a mostly pile front and


a leather back.

The next piece was the Tekke Turkmen mafrash below.


Although its devices repeat, they are densely placed and the piece has a fineness exceeding 500 knots per square inch.  For this latter reason alone, it is placed among pile pieces that were more difficult to weave.

Here is a closer detail of it.


I think much of the white is cotton.

Note: I think Marla would likely disagree that the densey arrayed design is a source of weaving difficulty but she might not object to the notion that at 500 kpsi, the weaver might have trouble seeing what she was doing without magnification.

The next piece was a non-Turkman, Central Asian, main carpet.   Its owner attribues it to the Uzbeks.


This piece had full pile and good color.  Rich Isaacson spoke to it, pointing out that it was a piece in which the weaver(s) repeatedly changed her/their mind(s) about the design.

To take only one more obvious example, the piece was begun with an entirely different border done in different colors than that ultimately adopted.


Despite the fact that relatively few design devices are employed rather spaciously, it is arguable that the weaver(s) found at least some of the designs in this rug difficult to weave.

The next piece brought in was a Baluch bag.


Here is a look at its back.

bi10aThe stripes of this back seem pretty straightforward but the density of the design and the employment of close analogous colors in it would likely make the pile face of this bag a middling difficult design to weave.

The next items shown was a comparision between two Baluch niche designs.


 David Hunt spoke to these two pieces,


saying that the piece on the right is antique, while the piece on the left is new.   They are both of a type described in the literature as products of the Djan Beki Baluch.  While there are some similarities, David said that the antique piece required more time and labor to make.  It has more colors, a finer weave, more complex end and side finishes, and a generally more complicated design.

David said that these two rugs illustrate well the changes that have taken place over time with regard to Baluch weaving practices and to those of carpet weaving, more generally. 

The last consideration of pieces in the room focused on the execution of striped borders in three pieces, but especially on that in one of them.

Michael Seidman spoke to this comparision.


He said that he wanted to compare how the weavers of these three pieces had planned for and handled the orientation of the striped main borders on them.

He started with the Afshar bag face below.


He said notice, in the image above, that the weaver chose to continue with the same rightward-leaning stripe, not just entirely across the bottom, but also up both sides and across the top.

Similarly in the Ladik niche design that John showed,


the weaver chose a different leaning, but also stayed with it all around the rug.

But in the the rug below, from the Baku area, often referred to as Khila, the weaver made different decisions with a view toward a particular outcome she wanted.

First, here is most of this latter rug as it hung on the front-of-the-room board.


Michael said that we should notice here that the weaver has slanted the stripes on the right side of the piece to the left and those on the right side to the right. 

More, examination of the top main border shows that the left side of it slants to the right and the right side of it slants to the left, and that these two sections meet in a perfect tent-like joint in the center.

In order to achieve this effect ,the weaver made decisions at the very beginning.  Michael asked that the bottom of the rug be held up so that we can see what moves this weaver made at the start.


Notice that as this weaver wove the striped main border at the bottom of this piece she oriented the stripe on the left side to the left and formed a joining “V” at the center so that she could continue weaving the right half of the bottom main border with its stripe leaning to the right.

BUT, when she got to the level of the bottom minor border at the edge of the field, she reoriented both stripes of the major border so that the one on the left side slants right and the one on the left side slants left.  By making this move she set things up, at this early stage so that she could have a top border that was a mirror image of the bottom one with a perfect joint at its center.

Michael offered this treatment as an instance in which an experienced weaver looked far ahead and moved skillfully to achieve a particular main border effect she desired.

John had placed a handout for this session on each chair as participants arose.

It gave a barebones summary of the lecture (something you don’t need here) and a list of sources.

Here is John’s sources listing:

Book: Humbert, Claude, Ornamental Design, New York, N.Y.: The Viking Press, 1970.

Bier, Carol, Symmetry and Pattern, a Textile Museum exhibition in 1997, an internet version of which is still on The Math Forum.

Email exchanges with the late Peter Collingwood, Marla Mallett and Mark Traxler.

If you want to see more of Wendel’s Swan’s treatment of Persian change purses, a salon featuring them is in the archives.  Here is the link:

To assuage his feelings of guilt for having violated nearly everything he knows about instructional design by having given a PowerPoint-assisted lecture nearly an hour in length, John included in the handout an exercise, experiential, like one Marla Mallett sometimes uses in workshop situations, but with very different objectives.

It is an exercise intended to let you experience a little of what a pile weaver does when selecting colors as he/she weaves.

Here are the directions for this exercise:

o  Examine the two images that follow these printed instructions.

    –  The top image is of the front side of a short section of a border on a Caucasian rug.  The design is a frequent version of the “trefoil.”

     –  The bottom image is of the back of this section.  It lets you see the knot more clearly.  Treat it, in this exercise, as if it is the cartoon from which you are weaving a pile rug.

o  Starting with the lower left corner of the bottom image, begin to mark (with a pencil) on the blank graph paper the color of each square in the cartoon.  The first square is “B” for brown and is the side selvedge.  After that, and moving to the right, is the color of a pile knot.  Since this is theh back and since knots go around two warps, both knot nodes are showing and each two of them comprise one actual knot.  So the next squares are yellow and there are six of them.  That’s three pile knots.  So record on your graph paper three “Y’s” in separate squares to the right of the “B” you have written for the selvedge.

o  The usual practice of pile weavers is to put all of the knots of a given color into a given row before moving to any other colors used in that row.  So, now you need to find where the next two yellow knots go (two squares to the right of the yellow knots you have already recorded).  Continue in this way until you hve put in all of the yellow knots in this first row.  Then go back to the left and put in the red knots, marking each appropriate square with an “R.”

o  The next row is easy, since it is mostly dark brown knots with only one red and one yellow one on the left side.  Put in that row.

o  Now the question is how far you want to continue this exercise?  There are some easy lower rows yet, but after you get to the trefoil designs, the counting and placement get a little trickier.  If you have the patience, try to continue with this exercise until you have, at least, completed the first trefoil device on the row with a red not in the trefoil’s center, you will have grappled, more than a little, with one part of what a pile weaver does as he/she weaves.

Expert pile weavers are reputed to be able to tie at the rate of 10,000 to 12,000 knots per day.   And they talk, constantly as they do it.

Seems miraculous, doesn’t it.

Well, it is said that many weavers are illerate an that they are similarly in awe of our ability to read quickly and steadily across one line after line of type without hesitation.

Below, the top image of the front and the bottom image is of the back of the border section referenced above.


I hope you have enjoyed this virtual version of this “rug morning,” despite its likely, tiring, length and convoluted character.

My thanks to Pat Riley for a good set of notes, to Thada Bornstein, who took the candid photos in the room, and especially to Marla Mallett who worked generously and patiently with me beforehand on this session and afterwards on this post.  It must be more than a little tiresome to have to explain (and sometimes re-explain) some of these aspects of weaving to a non-weaver.


R. John Howe

Samy Rabinovic: Color as a Tool to Identify Anatolian Carpets

Posted in Uncategorized on May 12, 2009 by rjohn

On January 17, 2009, Tom Goehner, the recently appointed Textile Museum Curator of Education, introduced


Samy Rabinovic, of Philadelphia,


as the speaker in a Textile Museum, Rug and Textile Appreciation Morning program on the subject “Color as a Tool to Identify Anatolian Carpets.”

Samy began by referencing Jon Thompson’s typology of the various types of oriental carpets, first published in his “Carpet Magic,” volume in 1983.

Samy said that, traditionally, rugs were classifed  primarily in terms of where they were made.   Kashan, Bokhara, Konya, Tabriz, Heriz, etc.

He said that Thompson proposed four groups of carpet groups based on the various contexts in which they were woven.


The four groups are:

Tribal, made primarily for the weavers’ own use and often woven by nomads.  In this category are such tribal groups as Lurs, Bakhtiaris, Afshars (Iranian), Kurds, Amenians, Lezghi’s (from the Caucasus), Yuruks; Anatolian Turkic tribes, plus  Kurds and Armenians; Central Asian tribes, including Turkmen, Uzbeks, Kirghiz, Karakalpaks, etc.

Cottage or Village, often with slightly more complicated designs, including more refined borders, to be sold in local  markets.  Here geographic designations still apply and include such locations as Heriz, Bijar, Hamadan in Iran and others, such as Bergama, Konya and Ladik in Anatolia.

City, commercial weavings.  Here the weaver is told what to weave.  Kashan, Isfahan, Qum and Nain, in Iran; and  Usak, Hereke, Panderma and Kumkapi, in Anatolia, are often cited as instances of this group.

Court Carpets, made in such places as India (Mughal), Turkey (Ottoman) and Iran (Safavid) courts.

Samy said that the carpets we would be treating in this session would mostly be those that would fit in Thompson’s cottage or village category, although some might be attributable to the tribal or nomadic group.

Samy is trained as a chemist and said that he began to look at color in rugs and textiles in 1984, influenced by that background, as well as by his acquaintance with Harald Bohmer, of DOBAG fame.  Samy said this perspective on color changed the way he looked at rugs, especially those from Anatolia.

He said that he came to see that colors were used, broadly speaking, in different ways in different parts of Turkey and that such color usages could be an important indicator of where a given piece was likely woven.

Still in his preliminaries in this presentation, he resorted to a color wheel presented in black and white, but with a key for distinguishing categories of colors (see Ref.1):


Here is a somewhat similar color wheel in color from the internet.

To summarize from the labeling in Samy’s “black-white” color graphic above:

Primary colors: red, blue, yellow

Secondary colors: green, violet, orange

Tertiary colors: red-brown, blue-green, yellow-brown

He said that the notion that there are noticeable patterns in the use of color by weavers of particular types or in particular areas is not a new idea.

Persian weavers in Nain often seem to favor  soft yellows and saphire blue.


Caucasian weavers use mainly primary colors.


The Baluch are famous for using mostly darker, analogous colors with white highlights.


And, of course, Turkmen weaving is a near synonym for “red rug.”


He said he thought that color usage also varied, rather systematically, by geography in Anatolia, and that he would attempt to illustrate and apply the color usage tendencies that he thought were visible and that could be used as an aid to attribution of Anatolian weavings.

He said that various proposals have been made about useful ways of dividing Anatolia for purposes of rug and textile analysis.

In a handout he provided this black and white map (Ref: Dr. Bohmer’s book, Rugs of Anatolia).


Here is a colored map that may be a little easier to make out here.


Samy  said that for the purpose of his color usage thesis he would define three broad areas.

As early as 1900 Mumford (and again in 1911, Lewis) classified Turkish carpets as Izmir (Western Anatolia) and Konya (Central Anatolia) carpets.

Today  we define Western Anatolian Carpet (See Ref 1: Rugs of Anatolia, by W. Bruggemann and H. Bohmer), the whole western part of Anatolia, starting with Canakkale, all the way south to Fethiye (famous for “Megri” carpets).

However, with a closer look  at the color distribution or color palette used in Western Anatolian carpets, it is easy to distinguish two areas.

The first, north of Izmir, starting with Bergama, all the way to Canakkale, with Ezine, Yuncu, and Balikesir in between. In this northern area, the primary colors are red and blue with minor amounts of yellow.

The second western area is south of Izmir.  Here we notice not only reds and blues, but also yellow, sometimes two shades of yellow, as is the case with Megri carpets; purple as in Cal (near Aydin), Milas and Ushak carpets.

There may be tribal influences visible in these color differences between these northern and southern areas of western Anatolia.

(An important comment, in this context, was made by Harold Keshishian from the audience.  Harold said that a collector friend of his referred to “Ada Milas” carpets of the 18th and 19th centuries as woven by Greeks.)  Even if carpets from this area were not woven by Greeks, the differences in color palette between the north and south western Anatolia do suggest different influences.

Central Anatolian rugs, the second area Samy designated, has been defined, he said, differently by different people.

He adopted a triangular area (it’s actually drawn on the black and white map above, the area defined by Bohmer et al,but may be hard to see). Defined on the west by an approximately vertical line from Eskisehir in the north to Mut, near the Mediterranian in the south.  Its northern boundary is an west-east line from Eskisehir through Ankara to Sivas.  An eastern boundary of this central Anatolian area is defined by a line from Sivas in the north to Mut in the south.  Konya is often thought as the “center” of central Anatolia, but it is actually on the western side of this triangle and of the actual center of Anatolia itself.

Other geographic locations in this Central Anatolia region are Karapinar, Mucur, Aksaray, Ladik, Gelveri, Incesu and Yahyali.

In Central Anatolia, all the primary colors are used, again with lots of yellow.  The secondary colors green and purple are also used.   Yellow grounds and yellow-purple combinations are frequent.  Borders in rugs of this area aresometimes multi-colored.

A very large, and admittedly disparate, Eastern Anotalian area is more difficult to define by predominant color usages.

Although marked by diversity,  Eastern Anatolian colors usages seem generally darker. 2-3 reds, browns and blues are noted, often as transition colors. Samy said that little yellow seems to be used in Eastern Anatolia (although there are exceptions).  And white is  often used to create compartment separations.

Geographic locations in Eastern Anatolia include Malatya, Gaziantep, Kagizman and Savak, just to name a few.

Samy turned, now, to apply this color usage typology to the pieces he had brought.

He started with this niche design from north-northwest Anatolia.


The use of blue (saturated indigo) and red is typical for theYagcibedir (near Balikesir) carpets of the upper area of western Anatolia.  Also, the use of white to outline the border of the mihrab, and the secondary use of orange on the outside borders ,is typical of Yagcibedir and Yuncu carpets.

His second piece is this Megri from southwest Anatolia, the Fethyie area.


Compared to the first example of western Anatolia (the Yagcibedir example above, or to Bergama or Ezine carpets) note the use of yellow and blue as the primary colors, together with some red and green. The Milas, Ada Milas , Megri carpets are typical examples.

It is interesting to question whether or not there are visible tribal influences in the color usages of the northern and southern regions of western Anatolia.


The rug below is the third piece Samy brought.  It was attributed to “Ada Milas” (the literal translation is “island Milas).


Samy’s next piece was attributed to Central Anatolia.


The next piece was a divan cover (note borders on three sides but fourth side is finished; it is not fragmented).


There is very little yellow in this piece.  Red and blues are varied.  The reds are from both madder and cochineal (cochineal is used in western Anatolia but very frequently in the East).  There is a darker and a lighter blue.  This divan cover is seen to have been woven in Eastern Anatolia and is definitely reminiscent of the Small Pattern Holbein carpets .

The next piece is a fragment of the group of rug described as “yellow-ground Konya village” rugs.


This piece does exhibit red, yellow and blue, the primary colors  and the ground is yellow.  But it also has the secondary colors orange, green and purple.  The white ground borders are “multi-colored with the colors of the field devices.

A very nice prayer carpet was submitted from the audience (see Dr. Boehmer’s Rugs of Anatolia Plate 40/pages 188/189


Central Anatolia.

The next rug Samy treated was this yastik.


Its owner said that he bought in part because of the unusually large amount of purple used in it.  He also said that Harald Bohmer, with it in his hands, estimated that all of the dyes in it, including the strong orange, are natural.  This piece is attributed to the Karapinar area of  Central Anatolia or further to the east.

The next piece was the Malatya kilim bag below.


The blue-ish charactor of its red suggests cochineal.  Cochineal red occurs in western Anatolian weavings but is much more frequent in those from eastern Anatolia.

The  next piece shown was this Kurdish Anatolia rug with a baklava design. Note the absence of yellow.  White is used effectively in the small borders that form the compartmented aspect of the design.  White dots brighten the coloration generally.


The next piece was thought likely to be western Anatolia, more precisely to the north of Izmir rather than the south, most likely a Bergama type.

Note to Samy: This piece has prominent yellow and greens and seems to have colors more like what we said rug woven below Izmir in the west have.  Why do we see it as a Bergama rug?

Note to John: Please consider that the yellow in South Western Anatolia (more or less a strong yellow that we see in Central Anatolia carpets), whereas in Northern Western Anatolia, the yellow is more a mixture of orange and yellow, as in the example shown…. Contrast the yellows in the Megri or Ada Milas… Also design wise the first and third border are often seen in Bergama type rugs…


The following piece was a classic Bergama rug estimated to have been woven about 1800.


The next rug was this one attributed to Karapinar.


An additional piece was this yatak woven by non-Kurdish nomads in the Konya area.


The next piece treated was likely from the Ladik area.


A next rug was this one.

Note to Samy:  We need an attribution on the rug below.


The next piece shown was this one probably from Kirsehir.


The next piece was attributed to Kirshehir.


The next piece shown was this small Western Anatolia, north of Izmir.


The following rug was this  Karapinar yastik.


The next piece shown was this Malatya kilim bag.


The next rug was this Central Anatolian yastik, woven  in Taskale, near Karaman, using synthetic dyes, around the 1920-1930’s.


The next weaving shown was this Central Anatolian yastik.


The next to the last piece shown was a HEYBE or donkey bag brom Western Anatolia (the one Samy is holding here).  It is a Yuncu.



This is a complete “heybe,” a Western  Anatolian saddlebag set.  Samy holds one face.  A long section of the striped back is showing.  The other face is on the opposite side of the striped area at the end furthest from the one in Samy’s hands.  The second of  the two images above shows that, as is typical, there is a connecting panel, longer than you would see on most Persian saddle bags,  and that it has a slit down its middle.

The last rug shown in Samy’s session was interesting because it shows Western characteristics, but it could be  a Bergama or maybe a Manastir type woven in Western Anatolia.


In summary, Samy said, his picture about discernible regional differences in color usage in Anatolian carpets is that:

In western Anatolian carpets , the dominant colors are red and blue for the upper part Bergama, Canakkale, Balikesir, Yuncu, Yagcibedir  and Ezine.  South of Bergama to Fethiye, we see all three primary colors red, blue and yellow.

In Central Anatolia, all of the primary colors are used as well as secondary ones like green and purple. The yellow ground Konya carpets (mostly from the Cappadocia area), and those from Karapinar, Karaman, Mucur, Aksaray, Ladik, Gelveri, Incesu, and Yahyali are good examples.

In Eastern Anatolia one needs to talk about the absence of rules and of a resulting diversity.  Carpets from this region are generally the darkest.  The use of  close “transitional” colors such as 2-3 reds or 2-3 blues, is common.  Yellow is generally absent, although there are exceptions, especially when camel hair is substituted with yellow dyed wool.  Rug producing areas include Malatya, Gaziantep, Kagizman, Shavak…

It is an interesting, imaginative attempt to use color as an aid to identifying the regions within which Anatolian carpets were likely woven. After all, the same raw materials (local plants) were available to weavers of the three different regions. Dyer’s weld, onion skin and 30 plus plants yielding the color yellow is known throughout Anatolia, yet North Western Anatolia (Bergama, Canakkale, Yagcibedir) and especially Eastern Anatolia, specially Kurdish rugs use very little yellow whereas Central Anatolia and South Western Anatolia seem to like the color yellow. One should ask the question why?

The session ended and folks came forward for the material and to ask Samy additional questions.


My thanks to Samy for his permission and assistance in producing this virtual version of his interesting RTAM program.

Thanks also to Wendel Swan, who provided a number of the pieces shown, and also helped in shaping the text of this post.

I hope you have enjoyed this virtual RTAM program.

R. John Howe