<---previous article


by Mark Hopkins.

This article originally appeared in HALI #93

Of all Islamic prayer rugs, only those woven by the so-called Baluch tribes define their mihrabs by abruptly altering the course of their borders. Mark Hopkins, inveterate Baluch rug collector and president of the New England Rug Society, examines a sample of two hundred such prayer rugs in an attempt to find diagnostic trends among the morass of different design permutations. The article contains unpublished pieces from the author's collection.

Figure 1. Baluch prayer rug (detail), Type 1b, 19th century. 0.81 x 1.47m (2'8" x 4'10"). The outer minor border continues upward to enclose the rug while the main and inner minor border bend inward to form the mihrab – the most common of all border configurations, found in 28% of the sample. Mark Hopkins Collection.

The apocryphal sage who said, "In their similarity, their dif­ferences are infinite," was characterising the entire human race, but he might just as well have been describing the weavings of the Baluch. Or perhaps, acknowledging conventional wisdom, we should say the 'Baluch'.

Unscrambling the Baluch omelette is a brave and worthy exer-

cise: many of us are no longer satisfied with the simple catch-all term and are intent upon progressive levels of subdivision. Is . this truly a Baluch prayer rug? Or is it a Timuri? A Bahluli? A Jamshidi? A Taimani? A Janbeghi? The assumption, of course, is that this group is essentially heterogeneous and that Baluch weavers did not intermingle; different designs were the products of particular 'sub-tribes'.

Figure 2. Baluch prayer rug, Type 1b, 19th ceutury. 0.84 x1.35m (2'9" x 4'5"). The widespread occurrence of mul­tipleadditional inner borders was ignored in the study inorder to simplify classification. Mark Hopkins Collection.

Despite great diversity in their details and graphic flourishes, 19th and 20th century Baluch prayer rugs comprise what is prob­ably the most consistently identifiable rug group there is. Even the most diffident beginner can declare with confidence, "This is a Baluch prayer rug." Why? Because the weavers of such prayer rugs do something instantly recognisable that the weavers of no other prayer rugs do: they define their prayer

arches (mihrabs) through a radical 90° inward bending of the main border system. 1

A quick visual scan of the best known Islamic prayer rug configurations confirms this. On all other principal types the mihrab is formed by employing design devices within a field that is framed by an undisturbed rectangular border system.

Figure 3. Baluch prayer rug, Type 2b, 19th century. 0.76 x1.22m (2'6" x 4'1"). The 'Turkmen' border is used in 35% of the total sample. It also appears on blue-groundTimuri' rugs, such as (6). Mark Hopkins Collection.

What caused Baluch prayer rugs to be so different? As every student of ethnography quickly grasps, change in tribal cultures happens at a slug's pace except in cataclysmic times of war and submission. What kind of ethnographic thunderbolt could have remapped the concept of rug borders for an entire generation of weavers and affected the shape of all Baluch prayer rugs to come?  Or is this simply an indigenous design emerging from the shadows of the past?

There are few clues. Parviz Tanavoli has published a curious Varamin weave gelim with the stepped shape of a namakdan, but of such a size that he identifies it as a prayer mat. 2 He suggests that such stepped mats might have formed the basis for the prayer arches on Baluch prayer rugs. But this is slim evidence; the former owner of the piece, John Wertime, reports this to be the sole occurrence of such a specimen in his wide experience.

Figure 4. . Baluch prayer rug, Type la, 19th century. 0.81 x l.32m (2'8" x 4'4"). The entire a-b-a border systems diverted inward to define the mihrab. Red-ground Baluch prayer rugs are rare, comprising only 3% of the sample. Private collection, London.

Even a cursory examination of a group of Baluch prayer rugs reveals an almost endless array of design variations within their unique format. One of the most interesting differences occurs in the border systems. Compare two apparently similar camel­-ground rugs: the borders of the first consist of a main border flanked by two reciprocal guard stripes, plus a 'Turkmen'-type border contained within (2). In this piece the mihrab is defined by

eflecting the main and inner guard borders (plus all the additional 'contained' borders) inward, while the outer guard border continues upward in the normal fashion to frame the hand panels and upper perimeter. The second prayer rug, by contrast, has a system consisting of two dissimilar borders, both of which bend inward to frame the mihrab, while the hand panels are separated from the selvedge by just three rows of knots 3.

Figure 5. . Baluch prayer rug, Type 1c, 19th century. 0.84 x l.52m (2'9" x 5'0"). The inner minor border defines the mihrab while the other two continue upward. Boucher Collection, Indianapolis Museum of Art.

It was casual observations of this sort that prompted me to examine a large sample of Baluch prayer rugs. The questions posed were basic ones: are patterns in mihrab configuration apparent that might allow these weavings to be grouped into a convenient, subdividable hierarchy? And if so, might that hierarchy shed any light on tribal and geographic origin?

The study was based on two hundred old or antique Baluch prayer rugs with variations of the bent-border mihrab.

Only those types of Baluch prayer rug that are clearly departures from the norm were excluded, specifically mosque-type prayer rugs, 'Dokhtor-e-Ghazi' prayer rugs, double-ended 'Aimaq' (funerary?) rugs, and panel-less 'Qara'i' prayer rugs. In other words, if it had hand panels and the bent-border configuration, and was not a new rug, it entered the study. My search through the literature suggested that more than nine in ten of published Baluch prayer rugs satisfied these criteria.

Baluch Prayer Rug, SW. Afghanistan, 19th century, 3’5” x 5’0” (104 cm x 152 cm)

In my humble opinion, the best prayer of the entire collection, mistakenly attributed to the Taimuri., a tribal name associated with weavers of NE Persia/NW Afghanistan. - T.Cole

 Of the pieces examined, 177 were illustrated in exhibition and sale catalogues, books and rug journals, a sample which appears to represent more than 90% of the Baluch prayer rugs published at the time. The other 23 were from photographs of unpublished rugs supplied by collectors and dealers, 18 of which I was also able to examine in person. Except in the case of these 18 rugs, technical analysis was based on published details.

   A computer database was compiled based on my observations. Its aim was to determine whether inter- relationships of common characteristics such as

border configuration, knot type, ground colour and design motifs might reveal any useful information.

The first thing to become apparent was that, while the border systems of these two hundred examples showed a startling variety, they did suggest that an approach to orderly grouping was possi­ble (see drawings). The simplest Baluch prayer rug can have just one border. Almost always, though, they are much more complex. The study included pieces with up to nine contiguous borders, not counting those rows of one or two knots that may separate the borders themselves. However, the vast majority of the rugs studied (72%) had three to five borders.

   Distinct groups became evident when the border systems were examined from outside in, focusing on what appeared to be the 'main' outer border system while disregarding all the 'supple­mental' wide and narrow borders contained within.

   The first major group, Type 1 (1,2,4,5,10) comprised 60% of the sample. The border system consists of a main border brack-eted by two identical minor borders (a-b-a). Type 2 (3, 6, 7, II) the second major group, encloses the rug within two different main borders (a-b). This group included 31% of the sample. Type 3 (8)

is a modification of Type 1 that adds a 'bridge' to connect the two hand panels across the top. While this group comprised only 4% of the total sample, it is noteworthy because its individual members include some of the older and more interesting Baluch prayer rugs on record. Two variations of the group which may be familiar are the regal piece on the cover of Adraskand's Belouch Prayer Rugs,3 and the Victoria & Albert Museum's prayer rug illustrated in Jenny Housego's Tribal Rugs.4 Types 4 through 7 and their variants made up the remaining 5% of the sample (9).

Figure 6. . 'Timuri' prayer rug, Type 2a, 19th century. 0.81 x 1.08m (2'8" x 3'61/2"). The inner border defines the mihrab, while the dissimilar outer border continues upwards. Blue-ground prayer rugs comprise 20% of the sample. Gerard Paquin Collection, Massachusetts.

What is interesting and potentially revealing is the multiplicity of ways in which the weavers of Type 1 and Type 2 rugs (91% of the sample) redirected their border systems to form their mihrabs. Each weaver, with her prayer rug on the loom, apparently arrived at the onset of the prayer arch with a host of options. From this point on, which of the borders were to be diverted inward? And which were to continue straight on?

As the schematic drawings demonstrate, the weavers of Type 1 and Type 2 prayer rugs enlisted virtually every possible permu­tation to form their mihrabs. Not only did they divert the borders in all conceivable ways, but they sometimes even ‘split’ them to frame the hand panels, as in Types 1d, 1e and 1f.

Baluch Prayer Rug, NE Persia, 19th century, 2’3” x 3’4” (69 cm x 101 cm)

Another lovely example, an older example of a design type that is very desirable in terms of the market as well as an aesthetically pleasing weaving of significant antiquity. - T. Cole

At this juncture, no consistent pattern had emerged from the findings. For every conceivable way to bend the borders while weaving a prayer rug, there clearly existed one or more Baluch weavers who favoured it. But can these varying border systems be related to other criteria by which a rug's tribal or geographic origins are sometimes identified, in order to see if any apparent correlations arise? I chose the obvious ones: knot type, ground colour and general design (unfortunately, the degree of warp depression was noted in only ten published examples).

My software could not come up with any meaningful correla-tions. More than half the sample (107 rugs) was identified in its original published source by knot type. Not surprisingly, 94% of this sub-sample was asymmetrically knotted. Of the 73 rugs where the direction of the knots could be identified, 92% were open left, 5% open right. The symmetrical knots of the remaining 3% were all pulled to the left. Technical analyses in three lead-ing books on Baluch weavings suggest that these figures mirror the norm for Baluch rugs, prayer and otherwise.5  Due to the preponderance of the asymmetric open-left knot type, it was clear that there was little to gain from an analysis in this area.

Baluch Prayer Rug, NE Persia, 19th century, ’1” x 3’6” (64 cm x 107 cm)

Seemingly more "archaic" than the previous weaving, but possibly from a later period as the drawing, while very spaciously rendered is not as clearly articulated nor as colorful - T. Cole

An examination of the sample by ground colour proved rather more rewarding. The 189 rugs actually seen, or illustrated in colour, had one of four ground colors: 75% camel; 20% blue; 3% ivory and 2% red. The rest were illustrated in black-and--white without mention of colour. As colour is believed to be an indicator of the origin of Baluch prayer rugs, one would hope to find correlations between specific border configurations and ground colour. Perhaps the blue-ground rugs popularly called 'Timuri' would show a

tendency to use a specific treatment of mihrab and border consistently different from that used on traditional camel-ground rugs of the standard Baluch type.

As is always the case with Baluch weavings, it was not that simple. All but one subtype of Types 1 and 2 were found among the camel- and blue-ground rugs.  If tribal convention required that Baluch border schemes were supposed to be different from those of their Timuri brethren, our weavers apparently hadn't heard about it.

Figure 7. . Baluch prayer rug, Type 2b, 19th century. 0.86 x 1.32m (2'10" x 4'4"). The mihrab is defined by both dominant borders, with only a thin guard border surrounding the rug. Camel-ground prayer rugs comprise 75% of the sample. Mark Hopkins Collection.

There were, however, some recognisable groupings: blue­-ground Type 1 rugs showed a tendency towards the Type 1a border configuration, while Type 1 camel-ground rugs were weighted towards the Type 1b configuration. Beyond that, however, the various border types appeared fairly evenly distributed among the two ground colours. 

Finally, we addressed the question of whether the various border configurations might reveal significant correlations with any of the most common Baluch border designs. Considering the almost infinite variety of such designs that appear in Baluch pile weavings, this caused my computer - and subsequently its operator - to convulse at the mere prospect of such a task.

It did, however, become clear that one particular border design was used in a full third of the prayer rugs studied, apparently occurring with greater frequency than any other readily identifi-able species of  border. This 'Turkmen' -type border (2, 3, 5, 6, 10), rare in non-prayer Baluch rugs, is also sometimes found in Salor and Tekke Turkmen weavings.  To my knowledge, the only attempt to assign a Baluch tribal origin to it was made by Jerry Anderson, who dubbed it "a Sangchuli idea".8

 Present in Type 1 (34%), Type 2 (42%), Type 3 (14%) and Type 4 (50%), the design was sometimes integrated into the main border system (10), while in other cases (3) it appeared as part of the supplemental inner border complex. It was found in all but one of the Type 1 and Type 2 sub-categories.

Figure 8. . Baluch prayer rug, Type 3a, 19th century. 1.14 x 1.49m (3'9" x 4'11"). The hand panels are connected by a 'bridge' over the top of the mihrab. This rare configuration (4% of the sample), is also found on camel- and red-ground rugs. Sold at Lefevre & Partners, London, 23 April 1982, lot 55. HALI Archive.

 The 'Turkmen' border appeared most frequently (45%) in camel--ground rugs - those most likely to be ascribed a truly Baluch origin - and was also evident in 11 % of the blue-ground rugs. Its ubiquity supports the idea that there was a widespread trading of border (and, for that matter, field) designs among the principal tribal groupings.  

 A half-hearted attempt was made to correlate age with the various groupings, but again with little success. There are so few valid benchmarks on which to base age estimates for Baluch rugs that we wasted little energy in this direction. The exercise did, however, suggest a reappraisal of one of the field's

minor conventional wisdoms: that the position of the hand panels in relation to the border can be an indicator of age.9 The survey showed a fairly even distribution of the different mihrab types among rugs of different generations, suggesting that this assump-tion may need reconsideration.         .

    So where does all this take us? The premise of the study was that since the bending of rug borders in new and different ways seems a rather radical departure for tribal weavers, it would follow that women of a given tribal group could be expected to adhere to a particular configuration. Clearly, however, that did not appear to be the case in our sample.

Figure 9. . Baluch prayer rug, Type 7, 19th century. 0.66 x 0.91m (2'2" x 3'0"). The single narrow border configuration occurs on only one of the two hundred prayer rugs in the initial study. This example was acquired aftert he study was completed. Mark Hopkins Collection.

It seems, instead, that our Baluch weavers demonstrated no great compulsion to restrict themselves to any strongly identifiable design convention. And judging from their older prayer rugs, this has not been a constraint for a century or more. The graphic content of their rugs seems to have been conceived in an exuber­antly creative, mix-and-match way.

This raises some interesting issues. Perhaps the weavers were more influenced by their own unfettered 'creativity' than by any tribal dictates. One also wonders how much of this creativity was guided by the dictates of a local market; indeed, it is curious that, to my knowledge, there is no evidence whatsoever of a Baluch person actually praying on a Baluch prayer rug 10

To approach the question from another angle, perhaps many of  these rugs were not made by tribal weavers at all, but by copyists producing them for local and/or export markets. There are those who would disagree, saying that no market for old Baluch rugs existed, but the sheer number of Baluch weavings that have made their way into European and American homes since well before the beginning of the 20th century quickly dispels that notion.

    The critical question remains, if so many different tribal groups actually did participate in the production of these rugs, why are their products all apparently so similar? The evident widespread exchange of design motifs in Baluch prayer rugs suggests a cultural intermingling of peoples governed by varying tribal heritages. Perhaps all these groups comprise a more tightly interwoven cultural confederation than we now believe.

Figure 10. Baluch prayer rug, Type 1b, 19th century. 0.78 x l.29m (2'7" x 4'3"). Here the weaver departed from the traditional mihrab form by adding extensions of the central border over the tops of the hand panels. Sold at Lefevre & Partners, London, 21 October 1983, lot 51.  HALI Archive.

    Recent evidence is mixed. Dietrich Wegner, who spent years ministering to the medical needs of the Baluch, has said that he never saw "any exchange of patterns or motifs as a result of close contact between any two groups". . 11  Jerry Anderson, a biologist who boasts decades of contact with these peoples, emphatically states that the tribes never copied designs from one another. 12 

But another observer with respectable field experience - Andy Hale - totally disagrees.13 And Bob Pittenger, who has examined extensive British intelligence reports from the late 19th century, recently wrote: "One of the most striking things... was the close proximity in which people of various tribes lived."14  He goes on to suggest that the likelihood of longstanding design traditions remaining intact in the face of such intermingling seems remote.


All of this is not to imply the futility of attempting to subdivide Baluch weavings as a route to more specific tribal identifications. Structure, especially knot type, remains a valuable criterion. The contributions of several writers describing symmetrically-knotted 'Aimaq' weavings, formerly called Baluch, have already been of value. Other weavers using the symmetrical knot - the Bahluli, for example - have also begun to receive attention. 15

    Here we have a representative sample of two hundred prayer rugs, with a vast diversity of designs, yet different from all other groups in the unique way their prayer arch is formed, upon which we are striving to pin labels that will precisely identify their tribal origin. Until further concrete information becomes available that ties specific designs, styles or techniques to particular subgroups, our current preoccupation with stringing out lists of tribal names appears premature. Perhaps we should focus on the far more inter­esting question of how such an apparent diversity of weavers, supposedly belonging to different but interrelated traditions, should produce rugs so alike as to defy further categorisation

Figure 11. Baluch prayer rug, Type 2b, 19th century. 0.89 x 1.09m (2’11”x 3'7"). Ivory-ground Baluch prayer rugs are rare; this example was one of just six found in the sample of two hundred rugs. Mark Hopkins Collection.


1. Certain other pile rugs from this area also occasionally show unusual configurations created by redirecting the main border system; stepped-top Turkmen door rugs, for instance. See, e.g.. Edelmann catalogue, New York, 24 October 1981, lot 75.

2. P. Tanavoli, Bread and Salt, Tehran 199, p.ll, shows a prayer mat measuring 0.71 x 1.02m (2'4" x 3'4").  Note that the captions of this and the adjoining Baluch prayer mat have been transposed.

3. M. Craycraft & A. Hali, Belouch Prayer Rugs, Point Reyes Station 1982, cover, p1.7.

4. J. Housego, Tribal Rugs, London 1978 pl.125, also cited in HALI 73,  pp.80-1.

5. S. Azadi, Carpets in the Baluch Tradition, Munich 1986; D. Black & C. Loveless, Rugs of the Wandering Baluchi, London 1976; J.W. Boucher, Baluchi Woven Treasures, Alexandria 1989.

6. The study disregarded ivory and red field rugs in this instance because the samples (5 red and 6 ivory) were too small to have statistical relevance.

7. See, e.g. L Mackie & J. Thompson, Turkmen: Tribal Carpets and Traditions, Washington DC 1980, pp.75,77,108,ll0,111,139.

8. T. Cole,'From the Horse's Mouth--Talking Baluch with Jerry Anderson', HALI 76,p.87.

9. Craycraft,op.cit.,pp.9,53; R. Pittenger,'Prayer Rugs of the Timuri and their Neighbours',in Papers, Presentations, 7th International Conference on Oriental Carpets, Dusseldorf 1996, pp.173,183.

10. Prof. Brian Spooner, a University of Pennsylvania anthropologist who has done extensive fieldwork with the Baluch, tells me he has never seen a 'Baluch' prayer rug used for prayer.

11. Diehr, F.M., Treasured Baluch Pieces From Private Collections, Wuppertal 1997, p.19.

12. Cole, op.cit, p.83.

13. A. Hale,’The Story Is Free', HALI 76, p.9l.

14. R. Pittenger,'Further Research on "Baluch" Rugs and their Weavers' unpublished paper, 8th ICOC, Philadelphia 1996.

15. D.Mallary,’Bahluli Rugs? Another Group of Symmetrically Knotted “Baluch” Rugs Defined', unpublished paper, 7th ICOC, Hamburg 1993.



I would have never reproduced this article on my own, all thanks are due to Frances Plunkett (Washington, DC) who scanned the images and offered the text to me in a Word document.

The approach that Mark has chosen in this article is a bit foreign to the way I think and view rugs, as I primarily focus on aesthetics and color, rather than the very structured, mathematical approach offered here.

Prayer rugs of the Baluch were, I believe from my own field experience, primarily woven for trade, for commerce rather than forming an integral component of daily life for a Baluch man and his family.  Few prayer rugs exhibit a wear pattern of use that would have been incurred during the ritual of repeated prayer.

Prayer was a very important component of daily life, and still is.  But the use of full pile prayer rugs was reserved, I believe, for the urban dwellers of some wealth, while mere straw mats or a blanket of some sort would serve the average ‘Baluch’ person quite well.

The "best of the collection" belonging to Mark Hopkins may be viewed on the New England Rug Society website (www.ne-rugsociety.org)

Copyright 1997 by Mark Hopkins Original text and photographs appeared in HALI July, 1997
With many thanks to Mark Hopkins for generously allowing me to reproduce the article as it originally appeared.