Home Page Discussion


Salon du Tapis d'Orient

The Salon du Tapis d'Orient is a moderated discussion group in the manner of the 19th century salon devoted to oriental rugs and textiles and all aspects of their appreciation. Please include your full name and e-mail address in your posting.


Weave Pattern: A Holistic Perspective on Rug Structure That May Be The Most Widely Used But Has Not "Caught On" With
Rug Collectors, Scholars And Writers.

by R. John Howe



In 1977 a South African couple, Ivan C. Neff and Carol V. Maggs, with apparent connections in the Washington, D.C. rug community, published A Dictionary of Oriental Rugs with a monograph on identification of rugs by weave.

The 130 page dictionary was first, but it is the monograph on "weave pattern" that has made this a distinctive book in the rug literature. Neff and Maggs did what seems not to have been done before and not often since: they argued that a holistic rather than an atomistic perspective on rug structure was appropriate.

In the 1970s rug analysis was still largely dominated by examinations of design and patterns visible on the front of rugs and part of what Neff and Maggs were arguing was that progress in attribution was more likely if one turned the rug over and looked at the back. This part of the advice that they and others were giving then has subsequently been adopted generally in scholarly rug literature and by the rug collecting community.

But Neff and Maggs went further. They argued that the most usual forms of structural analysis stood too close to the rugs being examined and were often exposed to mistakes because of this dysfunctional myopia. They noted that many rugs had structures that might seem very similar in close description but that which were obviously quite different if one stood back a little and looked at the "synthesis" of these individual characteristics at the level of "weave pattern."

"Weave pattern," they argued, was the "most stable" perspective from which to make judgments of attribution. They suggested that it was telling that the weavers who made them and the dealers who bought rugs, and who could distinguish them closely, hardly ever looked closely at structure. A quick glance at "weave pattern" was sufficient. They argued that something was being lost as rug analysis devoted itself almost entirely to close-up, even microscopic analysis.

Neff and Maggs clearly lost this argument, at least at the level of the subsequent rug literature. Most rug books that want to be taken seriously nowadays include a technical analysis of the structure pieces included. But almost nobody provides the kind of photos of the back of the pieces in their book that would permit readers to examine differences in weave pattern. The structural descriptions are those of the "close-up" school of rug analysis.

Only J. P. Willborg, the thoughtful Swedish dealer, has followed the Neff & Maggs lead, and then only in a small, but very nicely done catalog for a 1993 exhibition entitled Hamadan, that Tracy Davis Hartke mentioned in a post last week.

Neff & Maggs do not provide a tutorial on how to examine rugs from the weave pattern perspective, they simply do it for a variety of pieces. An examination of their descriptions shows that they look for:

1. The degree of vertical "ridging" visible in the warps. The more distinctive and the deeper the ridging, the more depression there is of alternative warps. A rug showing one and a half knot nodes alternating in its warps has alternate warps depressed about 45 degrees. Since both asymmetric and symmetric knots have two nodes, 90 degree warp depression is identified by looking for instances of isolated color on the back of a rug back that show only one knot node. This confirms that there is a second node of this color directly below it and so the alternate warps must be fully depressed.

2. Whether warps are visible as they pass over a line of wefts [surprise! usually but not always an indicator that a rug is "single-wefted" (or as Marla recommends the weft makes one "pick" as it travels across the rug)]

3. The number of wefts between rows of knots and the extent to which this number varies.

4. The color of the wefts.

5. The extent to which wefts are visible as they move horizontally across the rug.

6. The thickness of the weft as it moves across the rug and the extent to which this thickness varies. A "string of beads" appearance in the weft characterizes Heriz pieces.

7. Any visible pattern in the overall "look" of the weft (for example, some rugs from the eastern Caucasus exhibit a "ripple" effect in their wefts; Senneh weft lines are described as "wavy.").

8. The shape of the knot nodes, especially whether the height is greater than the width. This is a more difficult evaluation to make for someone just beginning to use weave pattern and Neff & Maggs are often irritatingly elliptical in their references to knot nodes shape. They will refer to a given knot node as "different" or "distinctive" often without indicating more specifically what they are alluding to.

9. The overall "look" of the weave taken more holistically. Here their descriptions use words like "fibrous," "stringy," "coarse," "uniform," "neat," "granular" and "refined."

Now one can see that there is some of this that is not in fact much different from what rug analysts, who do examine rug structure more minutely, look at. Degree of depression, number and color of wefts are routinely part of what most analysts of structure document. But Neff and Maggs don't refer much to knot count or knot ratios and they don't usually comment the nature of the materials (there is an occasional reference to cotton warps, or to wool) and they only rarely note such things a ply or spin at all. So they are, on balance, talking about something rather different under their rubric of "weave pattern."

Neff and Maggs say that confusion is possible as a result of close description between rugs such as Hamadans and Sennehs, although it seems to me only a real rank novice would mistake these rugs for each other and even then only if restricted to written descriptions and prevented from handling examples of them. Similarly, Neff and Maggs claim that rug weave shows that some distinctions often made, say between Daghestan, Kuba and Shirvan rugs are "quite untenable." They further claim that rug weave is a sound basis for recognizing "old Sarouks" and for distinguishing them from Sarouks woven more recently.

It seemed to me as I reread Neff and Maggs and reconsidered their claims, that "weave pattern" might be most useful as an aid in making closer distinctions. Here are two sets of examples. The first compares the weave pattern of Salor, Tekke, Yomud and Ersari pieces. The second compares four distinctive Hamadan weave patterns in the Willborg catalog.

In each case I will provide an image of the rug first, then a look at its back and finally the description that Neff and Maggs or that Willborg provide(s). Please note that Willborg also provides close technical descriptions as well and these are also included as a fourth item of information for each of his Hamadan pieces.

I notice that while he provides images of the back of each rug in his catalog, Willborg does not refer much to weave pattern in his explanations of his distinctions between different instances of Hamadan weaving.

Here are the four Turkman pieces in sequence.

"One warp is completely hidden behind its partner, a feature which we believe to be the norm in weaves of the Salor tribe of Turkestan. The nodes in the asymmetrical knots have a stringy appearance due to the loose ply of the pile threads - a feature not seen in the Saryk weaves. Another feature present in sections of all the Salor weaves examined by the authors is the barely visible extrusion of the thread of the concealed node, most clearly seen here in the central pink pattern. Wool and silk fragment: gift of Frank L. Loftus (acc. no. 08.376) by courtesy of Museum of Fine Arts, Boston."

"The neat parallel lines ofwarp pairs are a feature of this, the most common, Tekke weave pattern. We believe taht the Tekke were the most skilled of the nomad weavers and supposedly learnt the craft from the Salor. If that is so the available evidence indicates that they excelled their teachers. This magnificent carpet has a weave which is average for the group. The difference between this weave pattern and that of the Kizil Ayak is discussed in the caption to the Kizil Ayak illustration."

"The warps are all on an even plane but the warp lines are barely perceptible. Both nodes of the symmetrical knots are clearly visible, but except by following a colour on the transverse line, it is impossible to associate each pair of nodes. The weft line is seldom visible. The overall appearance is of avery flat and uniform pattern, and very different from the most common weave pattern associated with the Yomud which has a ridged appearance caused by the prominence of the warp lines."

"The use of an undyed weft which is visible along its entire length is a characteristic feature. On the vertical line, the nodes of the asymmetrical knots arfe longer than they are wide. A degree of depression of alternate warps appears to be a development of more recent times. The Ersari Afghan weave pattern has an overall coarser and more fibrous appearance than the weave patterns of other Turkoman rugs."

And here, courtesy of our friend, Vincent Keers, in The Netherlands, so that you can compare weave pattern differences more directly, the backs of these four Turkmen pieces, have been repeated side by side.

The key is: nm4 = Salor; nm7 = Tekke; nm10 = Yomut; nm13 = Ersari

It does seem to me that at the level of "weave pattern," the Salor, the Tekke and the Ersari are pretty readily distinguishable. And I think the word "fibrous" is a good one for describing the gestalt of an Ersari weave. For me, the Tekke and Yomud are harder to distinguish by weave pattern. On my initial reading I thought that Neff and Maggs were saying usefully that Tekke weaving had clear warp lines but that on Yomut pieces this was less definite, the wefts were hard to see and it was hard to tell which knot nodes were part of the same knot. (Tracy and I talked about this on the side last week in reference to the trapping she put up on the show and tell board.) But they are apparently saying that primarily about this particular Yomut ensi and in fact acknowledge that clear columns mark the warp lines of most Yomut weaving too. So, on their own reading, it seems often likely to be difficult to distinguish Yomut and Tekke pieces on the basis of weave pattern. And this is a place where such a distinction would be useful.

Please proceed to page 2.


Discussion