|
Design purity is most likely to be
maintained over long periods when motifs remain within a single weaving medium. Continuity
is especially predictable in the restrictive weaves. For example, designing in brocading
and warp-substitution, with their severe constraints, remains much more constant than in
knotted-pile or soumak. Any weaver can freely alter knotted or soumak motifs as she
pleases. It is design migration from medium to medium, however, that truly encourages
design changes and disintegration.
When a weaver attempts to use freely formed designs (knotted, embroidered or soumak
designs) in a technique with more constraints (such as brocading, warp substitution or
slit tapestry), the task is often impossible. Major alterations are usually required. Such
borrowed forms are often difficult to integrate and may survive only briefly in the new
medium. When we consider the overall picture, in Middle Eastern village and nomad weaving,
the flow of design influence in this reverse direction has been minimal.
On the other hand, where influence flows more normally -- from restrictive techniques to
freer ones -- design adaptation is rarely necessary. Direct copies are possible, as we saw
earlier in the soumak copy (Figure 2) of a zili brocade design (Figure 1). Changes are not
forced. Modifications can be easily made to suit each weaver's fancy. Changes are most
often minor and gradual. I would like to show a few of the ways that designs are routinely
modified. |
|
|
First, structural design devices may be distorted when their raison d'étre no longer
exists. The common border of kilim hooks in the slit tapestry here has been transformed by
a Kazak knotted-pile weaver (far right). The horizontal bars, used to stabilize the weave
in tapestry but redundant in knotted pile, have been misplaced and a new design has
emerged. This form then became institutionalized, and variations on it appeared in several
places.
A similar kind of distortion occurs when knotted-pile weavers copy standard kilim border
crenellations, but turn part of their crenellations vertically. [25] |
|
 |
|
Figure 22a.
Slit-tapestry border design with horizontal stabilizing elements. Anatolia.
Figure 22b. Knotted-pile border with slit-tapestry design, but structural
elements altered. Kazak, south Caucasus.
|
|
|
 |
Second, color
distribution and usage may be altered in a new medium, where its placement is not dictated
by structural requirements. When a carpet weaver borrows a warp-pattern design such as the
common arrow and half-arrow jajim motif, she is not limited to two colors in vertical
succession. In the first Caucasian border below, several color changes have been used to
visually help join two offset rows of arrow halves. Outlining adds one more color element. |
|
 |
|
 |
|
|
|
Figure 23a, b,
and c. Caucasian knotted-pile borders showing adaptations of the common warp-
substitution half-arrow motif.
|
|
|
|
Third, critical design
proportions may be altered in a new medium. Earlier we saw a Tekke knotted ensi border
(Figure 9a) that used the features of the "ainak" warp-substitution design, but
altered the spacing so critical in that technique. The knotted-pile weaver, with no warp
tension problems to worry about, simply squeezed the motifs together.
Working with few technical limitations, the knotted-pile weaver is also free to reshape
motifs. The first border above shows a fairly straight-forward copy of the basic
warp-substitution motif. Spindly linear elements have been used to join two separate,
multi-colored borders of arrow halves. A strong, reciprocal field/ground design
relationship has been retained; the white background itself makes a forceful
pattern. In the second border above, the vertical sequences have been severed, the
hooks have been reversed, and the diagonal joinings have been handled perhaps more
successfully; the white background, however, has been shattered. The weaver of the third
border has gone still further in an attempt to integrate the separate parts: she has
curved both the hooks and triangular bases, then turned the vertical arrow shafts
inward. Refined versions of this design are far removed indeed from the simple,
precisely engineered warp-pattern archetype. |
|
|
|
The next set of borders, below,
shows the gradual disintegration of another archetypal warp-patterned motif -- the
double-weave band design that we saw previously in a Beyshehir carpet border (on the left
here). In the second border, the blocks have been separated and variously colored. In the
third example, a Star Kazak version, color changes dissect the simplified border blocks
themselves, while in the last, on the right, the motif's framing has disappeared
entirely. Such drastic changes in color placement and design can easily occur in
knotted pile because the restraints of the original warp-pattern technique no longer
apply. |
|
 |
 |
 |
 |
|
|
|
Figure 24a, b,
c, d. Knotted-pile borders showing evolutionary changes in one archetypal
double-weave design. The third border is from a Caucasian Kazak rug; the others are
Anatolian.
|
|
|
|
A fourth kind of design disintegration involves a shift of focus from positive to negative
design elements. The common warp-pattern "S" border has begun to disintegrate in
this Bergama knotted-pile rug because parts of the background have been variously colored.
The Anatolian carpet below shows how readily transformations occur when the original
warp-pattern color limitations no longer apply. Common warp-pattern arrows in the vertical
border have been turned sideways. Spaces between these brown arrows have been
outlined and colored variously, transforming them into positive shapes. In the horizontal
border, at the top, these new forms have been alternately flipped, so the dark brown
arrows disappear entirely. |
Figure 25. Knotted-pile carpet border. Bergama area, western Anatolia.
|
|
Figure 26. Knotted-pile carpet with borders showing the shift of focus from positive to
negative design areas. Central Anatolia.
Courtesy, Harald Böhmer.
|
|
|
|
Fifth, when
motifs migrate, designs are often simplified, segmented and isolated. Cohesiveness may be
lost, as positive/negative design relationships are altered. The familiar
"kotchak" cross in the knotted Tekke border here can be viewed in two
ways: as four diagonal arms with hooks, or as four inward-pointing arrows. The weaver
must see both. In the second photo, a dramatically simplified Lori/Bakhtiari soumak
version, only four arrows have survived; the linear, hooked arms have disappeared.
Everything is disjointed; the simple background shapes have even become small, isolated,
stepped figures. When this design was transferred from one structure to another, a new
mind set was required of the weaver. [26] Cohesion and strong positive/negative design relationships are among the
most useful of all clues when tracing design origins. |
|
 |
|
Figure
27a. Tekke knotted 'kotchak' design of four arrows.
|
|
 |
|
Figure 27b.
Lori/Bakhtiari soumak border with negative
design elements isolated to become separate figures.
Courtesy, James Opie.
|
|
|
When motifs migrate WITHOUT
changing form, to appear as either major design elements or minor details in a less
restrictive medium, they may be quite different stylistically from other parts of the
design. We previously saw contrasting styles of pattern execution that occurred when
warp-patterned designs migrated to knotted pile and soumak. The soumak bag at the right,
as well as Figures 10 and 12, show motifs with stepped diagonals in borders surrounding
fields with smoothly executed design elements.
Likewise, the saddlebag we saw earlier shows marked contrasts between parts of its design
that have different origins. Although the field has a blocky, stepped zili brocade
design, the large border in this weaving uses tapestry motifs with smooth-edged diagonals.
The contrast is subtle and effective. [27]
Stylistic inconsistencies within complex field designs are equally informative. They
should alert us immediately to the likelihood of design migration. Within a single field
pattern, the presence of stepped motifs along with others that are smooth-edged can be a
telling incongruity. But numerous, less obvious stylistic inconsistencies can lead us to
conclude that parts of a pattern originated elsewhere. We may find further discrepancies
in the construction or proportioning of designs, peculiar contrasts in scale, varying
methods of pattern articulation, diverse approaches to the use of negative space, or
conflicting concepts of pattern organization. We can even find instances in which the
weaver has made mid-course alterations in weave structure to accommodate borrowed design
forms. These overlapping and multifaceted subjects are too complex to be covered
here. |
|
|
|
|
Although we may discover elements of
borrowed design in nearly every fiber medium, incongruous or unassimilated alien forms
appear the most frequently, by far, in knotted-pile rugs, soumak textiles and embroideries
-- the least restrictive structures. I would like to comment briefly on one case in point
which is simple and clear enough to require no structural or technical explanations: an
Azerbaijani embroidery with major elements derived from knotted-pile carpet design.
Although several writers have pointed out similarities between embroideries and carpets
from the Southern Caucasus, I have read no satisfactory arguments to support their
speculations that embroidery designs inspired the carpets. [28]
In my opinion, the embroideries are nearly always the copies.
A detail of a large Azerbaijani blossom carpet from the Iparmüveszéti Museum, Budapest,
is shown below. In HALI 59, Jennifer Wearden contrasted this carpet design with
that in with a small domestic embroidery. [29] The parallels between these two pieces are striking: the major
motifs are nearly identical. But the differences are dramatic as well.
In the large workshop carpet (below), minor motifs are consistent with the main figures.
Small, angular leaves just above and below geometrically- shaped medallions are repeated
throughout the carpet; the angular forms are echoed in background tracery. This knotted
carpet is formal, angular and symmetrical throughout. |
|
|
|
Caucasian Floral Carpet. Detail.
Iparmüveszéti Museum, Budapest, 24.462
|
|
The small embroidery has no
such consistency. Its central medallions and pendants are formal, symmetrical, and rigid,
just like those in the carpet. These forms, with their constant 45-degree diagonals and
their horizontal/vertical orientation, are natural products of coarse, knotted-pile
weaving. A few angular leaf forms are also present in the needlework. Most of the
embroidered background motifs, however, are sprawling forms that twist and turn to fill
the spaces. They nearly overpower the main design. The smallest details are inconsistent
as well. A peculiar mix of curvilinear appendages sprouts from the secondary medallions in
the center row.
This embroidery, with its naiveté and charm, is a hodge-podge of
forms: free-wheeling exuberance is mixed with the staid angularity of pseudo woven
shapes. Two completely different esthetics have been combined. Even the embroidery's
narrow border is composed of motifs that are structurally generated in woven
form: they are small, rather weak, serrated medallions. The embroidery's
numerous inconsistencies tell us that this needleworker copied standard rug motifs, then
filled in her background freely, in a comfortable, familiar and unrestrained manner.
Conversely, not a single embroidery characteristic appears in the large, formal workshop
carpet. The carpet's execution shows no attempt to circumvent the natural limitations or
inclinations of its technique. [30] Over the years, in this Caucasian carpet tradition, lancet leaves and
palmettes, both with angular interior details, gradually changed from crisp, cohesive
forms to an assortment of bulges, angles and protrusions. It is completely illogical that
an embroiderer, working with no loom constraints should have devised, by coincidence, this
particular woven convention -- especially a degenerate woven convention that
represents merely one stage in a long evolutionary sequence. The various elements in this
design are easily traced to early dragon rugs and even earlier Persian floral carpets.
But we need hardly be aware of this history to identify the needlework as a copy. Not only
are the differences between the embroidery and the knotted carpet informative,
the similarities are telling. Since the nearly identical motifs in the two
pieces reflect basic knotted-pile design limitations, and minor motifs in the embroidery
transcend such limitations, our basic premise applies: design influence has flowed in the
normal direction, from the more restrictive to the freer medium. |
Azerbaijani Embroidery. Detail, HALI
48, cover. Textile Museum, 2.18
|
|
|
Variations on One Design
Element:
The Anatomy of a Hook
|
|
| I would like now to follow a single detail
as it travels about from one textile medium to another. I would like to examine the anatomy
of a hook. |
|
|
|
In knotted pile, the hook is most naturally
an angular, horizontal and vertical device, although its outer corners can be easily
rounded. This Memling gul illustrates both forms. The most naturally evolving hooked form
in decorated reed screens is also the simple horizontal and vertical version.
Knotted hooks can be refined or elaborated in a number of ways. A diagonal stem can be
added to make the form seem more curvilinear. The all-over hooked patterns in Aimak
(Mushwani) carpets use this kind of form. [31]
If, on a larger scale, the interior corners are rounded as well,
the form becomes octagonal. |
Figure 28. Two kinds of knotted-pile
hooks: rectangular and rounded. Central Anatolia.
|
|
|
|
|
|
| Points may be added
to the top, bottom or sides of knotted-pile hooks, while the interior, basic structure
remains horizontal and vertical. Some Caucasian rugs exemplify this approach, as do a
great many Turkmen weavings. |
|
Figure
29. Rectangular knotted-pile hooks with points added. The Caucasus.
|
|
|
|
Figure 30. Diagonal brocade hooks. Kurdish
cover, Anatolia.
|
In contrast, brocade hooks are
normally diagonal, formed of progressively offset weft floats. They are angular and
linear, with the parts neatly parallel.
This diagonal form is sometimes duplicated in knotted pile. These hooks often appear in
Baluch knotted-pile borders, as well as in one type of Yomud asmalyk field (below). We see
brocade-type hooks in the border figures of the early Anatolian animal carpet in the
Metropolitan Museum. [32]
This kind of hook sometimes becomes an angular spiral. |
|
|
Figure 31. Brocade-type hooks in knotted
pile. Yomud asmalyk.
|
|
| None of these forms is
satisfactory for slit tapestry. The first, the vertical and horizontal knotted-pile hook,
requires long slits. The second, the brocade hook, with parallel diagonals, makes an
extremely weak slit-tapestry structure. The tapestry weaver's solution is ingenious: she
simply adds a stabilizing horizontal element to the basic diagonal brocade form by filling
in the crook to make a triangular-ended form. This way it is strong. This hook is
everywhere in kilims -- always diagonal and horizontal. |
Figure 32. Medallions with reciprocal
slit-tapestry hooks. Konya kilim, Central Anatolia
|
 |
|
|
Standard kilim
hooks can be flattened or rounded on top and still maintain their structural integrity.
When kilim hooks are elaborated to become spirals, they are often flattened. This
emphasizes their horizontal and diagonal orientation and makes a stronger construction. |
Figure 33.
Flattened slit-tapestry hooks. Kilim from the Burdur area of Western Anatolia.
|
|
|
 |
When hooks are used to edge a
kilim medallion, the wefts forming the stem of each hook cross into the medallion or
diagonal border itself, significantly strengthening an otherwise long, weak diagonal edge.
Such kilim hooks are more than pure ornamentation. When we ponder the origins of the
common hooked medallion, we must consider this important structural function. |
|
 |
|
| When we see kilim hooks copied in
knotted-pile rugs, this detail provides useful information on the carpet's tradition.
Kilim-type hooks, for example, are prominent in the early Berlin "Dragon and
Phoenix" carpet, and in the Metropolitan Museum's early Anatolian animal rug. In the
"Marby" carpet (shown here) we see generic kilim hooks, but also awkward
attempts to adapt these hooks to vertical and horizontal medallion edges. This is most
unnatural in slit tapestry, where diagonal hooks are constructed only to ornament diagonal
edges. |
 |
|
|
Rarely are graceful, powerful and precisely
drawn triangular hooks well articulated in media other than tapestry. Fuzzy pile doesn't
do them as well, unless the hooks are very large. Jaf Kurds have used offset knotting to
handle triangular hooks somewhat better, but not with total success. [33] Knotted-pile examples that combine
various types of hooks, such as this Anatolian rug, typically display strong,
well-articulated squared hooks, but weak and wobbly diagonal kilim hooks.
Likewise, soumak textiles often fail to articulate triangular kilim hooks gracefully. This
is especially true of Lori/Bakhtiari and Kurd soumak examples [34],
but also applies to Shahsevan work. Nor are triangular kilim hooks well
suited to the large wrapped units of decorated reed screens (below). We must remember that
articulation is an important clue to a motif's origin. Clarity is important. If a form is
consistently clumsy, it probably originated elsewhere. |
Figure 34. Knotted-pile carpet with both
rectangular hooks and triangular kilim-type hooks. Central Anatolia.
Courtesy, Harald Böhmer.
|
|
|
Figure 35. Kurdish reed screen with
triangular kilim-type hooks. Eastern Anatolia.
Courtesy, Josephine Powell.
|
|
|
In assimilating triangular
kilim hooks, the weaver of this weft-substitution rug turned half of her red hooks
sideways -- those at the top and bottom of the central motif. Knotted-pile carpet weavers
sometimes do the same, since tapestry restraints do not apply.
The Seljuk carpet below has triangular kilim hooks fitted neatly into the side border
design. But the top border is a mess. This weaver seems to have been comfortable with the
ordinary kilim form, but could not handle the vertical adaptation she truly needed when
her knotted kufic border changed directions at the top. |
Figure 36. Weft-substitution weave with
half of the triangular kilim-type hooks turned sideways. Afshar, South Persia.
|
|
|
Figure 37. Konya knotted-pile carpet.
Kilim-type hooks are featured in the kufic border; they change directions unsuccessfully
in the end border at the top.
Turk ve Islam Eserleri Muzesi, Istanbul, Inv. no. 681.
|
|
| The weaver of the Bergama carpet below has
used traditional pairs of small rounded pile-carpet hooks to ornament the top, bottom and
sides of her medallions. On the diagonal edges she has substituted diagonal, triangular
kilim hooks. We can understand this substitution if we look inside the medallions at the
radiating arms. There, on the diagonal, she had considerable difficulty in handling the
usual carpet forms. The substitution of flatweave forms for standard knotted-pile carpet
conventions, the appropriateness of such forms, and the degree to which they are
integrated, are all useful clues in tracing a tradition's development. |
|
|
Figure
38. Bergama knotted-pile carpet showing the substitution of kilim hooks for standard
carpet forms on diagonal medallion edges. Western Anatolia.
|
|
|
| Most importantly, though, when we follow our
hook from medium to medium, we must be aware that RECIPROCAL designs are much more
likely to evolve in certain weaves than in others. The freedom inherent in knotting and
soumak neither fosters nor discourages reciprocity. In more restrictive weaves, however,
reciprocity is either a natural characteristic or is encouraged to varying degrees. In one
of the most common kinds of Anatolian brocading, for example, three-span floats alternate
on a fabric's front and back sides, automatically forming designs with reciprocal elements
(Figure 30). In warp substitution, warp tension problems encourage reciprocity. In these
cases reciprocity is structurally generated. |
|
|
On the other hand, the reciprocal designs
characteristic of slit tapestry are encouraged by the unique weaving process. A
slit-tapestry artisan need not complete rows of weaving across a kilim's entire width.
Instead, she can weave individual sections separately. There is only one restriction:
sections below diagonals must be woven before anything above those
diagonals. This sections of the background often must be woven before positive parts of
the design. Everything is simplified if the background consists of well understood shapes.
Because of this, motifs with identical but reversed positive and negative shapes --
interlocking hooks of the same size and the same shape -- develop naturally in slit
tapestry. The weaver of the kilim shown on the loom is constructing identical,
interlocking red and black hooks.
The reciprocal design in the Konya kilim of Figure 32 (detail below) is truly basic--it is
technique generated. Similar hooked forms are found in slit tapestry around the
world. The Peruvian slit-tapestry tunic below, from the 11th century, has the same forms. |
Figure 39. Slit-tapestry kilim on the
loom, with reciprocal hooks being woven. Fethiye area of Southwestern Anatolia.
|
|
|
|
|
|
|
Figure 40. Reciprocal
slit-tapestry hooks of archetypal form in an Ica/Chincha tunic. Peru, c. 1000 A.D.
Courtesy, Paul Hughes.
|
|
|
Reciprocal hooked designs
certainly can be copied in knotted pile, soumak or even embroidery. But maintaining such
forms over a long period, when they are not inherently natural, is another matter. The
critical proportions are easily changed; reciprocity is easily lost. In the Konya pile
carpets below, we see the typical progression: on the left, brocade-type hooks edging the
medallions are reciprocal. In a later carpet, on the right, the hooks have become merely
stuck-on, sprawling curls. They are not integral parts. |
|
 |
|
|
Figure 41.
Konya knotted-pile carpet with reciprocal brocade-type hooks lining the medallions.
Central Turkey. Courtesy, Sotheby's.
Figure 42. Konya knotted-pile carpet with non-reciprocal hooks lining the
medallions. Central Turkey. Courtesy, Harald Böhmer.
|
|
|
| Most "dyrnak" guls in
Turkmen carpets feature both knotted-pile hooks and triangular kilim hooks. In the Göklan
example on the left, the kilim hooks are reciprocal; in the Ersari example,
reciprocity has been lost. This kind of change -- the loss of reciprocity -- occurs much
less often in slit tapestry, because the tapestry process exerts the same design pressures
on modern weavers as it did on weavers centuries ago. |
|
 |
 |
|
Figure 43.
Göklan knotted-pile carpet with guls formed of reciprocal hooks.
Figure 44. Ersari knotted-pile juval showing the loss of reciprocity in the hooked
dyrnak guls.
|
|
|
|
In tapestry, when
hooks become decorative space fillers, stuck-on flourishes, or sprawling totems, the
motifs may be exotic or dramatic, they may have symbolic meaning, but they are not truly
archetypal forms. They are mutant offspring.
Likewise, in soumak variations, when skinny, non-reciprocal triangular hooks become
irregular bird or animal-head attachments, as in Lori/Bakhtiari soumak work, they stray
far from the archetype. It is my opinion that the purest triangular hooked forms -- the
prototypes -- are reciprocal forms, rooted in slit-tapestry production. Knotted-pile and
soumak versions are copies or adaptations, not always degenerate, but certainly
derivative. |
Figure 45.
Kuba slit- tapestry kilim showing a loss of reciprocity in the hooked elements. Caucasus
|
|
Figure 46.
Lori/Bakhtiari soumak "animal-head" column composed of
diagonal/triangular hooked forms far removed from reciprocal slit-tapestry medallion
archetypes.
Courtesy, James Opie.
|
|
|
|
Design influence flows normally from
restrictive to less restrictive techniques. Design features that are dictated by
structural limitations clearly indicate their origins. Structural problems should
alert us to outside influences. Inconsistencies in design execution often indicate diverse
design sources. Fine articulation, cohesiveness and strong positive/negative design
relationships can point us toward likely origins. And last, design change and
disintegration is accelerated as motifs migrate from medium to medium, since different
technical constraints apply.
Studies of design origin have often been characterized by baseless theorizing,
unsubstantiated claims, and wishful thinking. Work in this field must, at the very least,
be based on an understanding of the following questions:
1. Can any single design be produced, without changes, in every fiber medium?
The answer: No. Several structures and techniques are so restrictive that they
prevent the use of certain forms without substantial alteration.
2. Do designs exist that are technique dependent -- that can be produced in only one
medium?
The answer: No. The freest techniques -- soumak, knotting and embroidery -- can
copy almost anything. More restrictive structures permit limited design exchanges.
3. Are certain designs typical of particular techniques or structures? Are there
truly technique-generated designs?
The answer: Yes, definitely. With every weave structure and process, a
distinctive repertoire of naturally evolving forms is generated. It is in the medium of
its origin that each pure, archetypal form is found. |
|
|
|
|
|
|
25.
|
A related
kind of alteration is apparent in a knotted rug illustrated by James Opie, op. cit.,
Plate 14.14. Although this rug has vertically banded patterning, it is not, as Opie
suggests, a copy of a warp-patterned jajim design. Instead, it shows weft-substitution
patterning turned on end. Although this knotted example closely copies a familiar
flatweave product, it is easy to imagine how subsequent changes might effectively add to
the confusion concerning its origin.
|
|
26.
|
Soumak
wrapping is slightly more continuous, linear and horizontal in its emphasis than knotted
pile. The soumak weaver is more likely to separate ground and figure -- to
"in-fill" a background -- than is a knotted-pile weaver. Positive and negative
forms are less often given equal attention.
|
|
27.
|
See
M. Mallett, 1993, for a discussion of ways that an unsuitable weave balance, along with
contrasting styles of patterning execution, can indicate pattern borrowings.
|
|
28.
|
See
Jennifer Wearden, "A Synthesis of Contrasts," HALI 59, pp. 103-111. See
also, Carol Bier, "Weavings from the Caucasus: Tradition and Technology," HALI
48, pp. 16-25. Provocative articles by Christine Klose have been concerned with a limited
group of embroideries that raises a separate set of questions.
|
|
29.
|
The
Textile Museum embroidery is also published as the cover photo of HALI 48. This
same embroidery appears as Fig. 9 (p. 17) in C.G. Ellis, Early Caucasian Rugs,
Washington, 1975, where it can be compared with a closely related carpet in Plate 26.
|
|
30.
|
Embroidery
characteristics occasionally discernable in weavings, along with the woven features that
often appear in embroideries, make up a complex subject in itself. It is the
striking inconsistencies that are of primary importance in the current embroidery example.
We find parallel kinds of design inconsistencies in village knotted-pile carpets that
include motifs borrowed from brocading, warp substitution, tapestry, etc.
|
|
31.
|
See
Opie, op. cit., Plate 13.21 or Jeff Boucher, Baluchi Treasures,
Alexandria, Virginia, 1989, Plates 62 and 63.
|
|
32.
|
See HALI
53, p. 154.
|
|
33.
|
Opie, op
cit., Plate 9.9.
|
|
34.
|
Opie, op.
cit., Plates 5.37 and 8.6.
|
|
|
|
©
Marla Mallett, 2000
|
