Subject |
: |
Structure-limited designs |

Author |
: |
Steve Price |

Date |
: |
11-23-1999 on 02:11 p.m. |

sprice@hsc.vcu.edu Dear People, Daniel speaks of structure-limited designs in his Salon, and it occurs to me that there are probably some readers who don't know what this means. So, at the risk of sounding pedantic, I'll be pedantic. The simplest textile structure is a plainweave; warps running the length of the textile, wefts running the width, with wefts weaving between the warps. If wefts are heavy and packed down tightly the warp will be hidden and we will see only weft. That's a weft-faced plainweave, and the only kind of design that can be woven into it is horizontal stripes. If wefts are thin and packed loosely we will see only the warp. That's a warp-faced plainweave, and the only kind of design that can be woven into it is vertical stripes. If wefts are of about the same thicknes as warps and packed about as densely as the warps, we will see both warps and wefts. That's a balanced plainweave, and the weaver can make horizontal stripes, vertical stripes, and, since those can intersect, checks and plaids. All of these designs are structure-limited. Let's go a step further, and make a weft-faced textile in which the individual wefts don't each go across the full width, but go partway across and another weft of another color enters the width at the warp after the last one the first weft covered. This is a form of tapestry. Obviously, the weaver can now make patterns more complicated than simple horizontal stripes. But suppose she tries to make vertical stripes. The adjacent warps where the color changes are not connected to each other, so if she makes the vertical stripes run for more than a very short distance the fabric will have large slits and will be unstable. That's why this is called slit tapestry. The way to make it stable and still have some interesting designs and motifs is to use lots of diagonal color changes. That way, the unconnected parts of adjacent warps are very short, and the textile is stable. This is a reasonable explanation (first proposed, to the best of my knowledge, by Marla Mallett) for the very widespread occurrence of diagonals and motifs like latchhooks on slit kilims. The design was structure-limited or, if you prefer, structure-dictated. Since tapestry weaving almost certainly predated pile weaving, designs dictated by structure of tapestry were probably carried over into pile textiles, even though the structural limitations don't exist in pile. There are alternative explanations for the occurrence of these designs in pile rugs (latchhooks can be interpreted as animal heads, for example), and the two explanations are not mutually exclusive. I hope this is helpful to some of our readers. Pedantically, Steve Price |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Patrick+Weiler |

Date |
: |
11-23-1999 on 11:21 p.m. |

jpweiler@worldnet.att.net Steve, At the risk of sounding didactic, I will be didactic. You stated: "designs dictated by structure of tapestry were probably carried over into pile textiles". This is probably a valid statement; one relevant to the topic of the Ala Chuval being discussed. The preceding conclusion in the same sentence, "Since tapestry weaving almost certainly predated pile weaving" could be disputed on grounds of there being no evidence to confirm it. I am not saying it is not true, but the assumption you infer is that because flatweaves predated pile, designs from those weaves carried over into pile weaving. However, the implicit conclusion one can draw from this line of reasoning is that, therefore, pile weavings post dated flatweaves (because pile weave designs did not carry over into flatweaves). Your earlier paragraphs, though, state that flatweaves have inherent design limitations; therefore, even if pile weaving predated flatweaves, pile designs that are incapable of being reproduced in flatweaves could not have been carried over into flatweaves. (The lack of a body does not mean a murder has been committed.) This invalidates your conclusion that flatweaves predated pile weaves since there is "evidence" of flatweave designs being "copied" in pile weaves.. This leads us to the question: Which came first, the chicken heads or the latch hooks? Confusingly, Patrick Weiler |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Steve Price |

Date |
: |
11-24-1999 on 06:38 a.m. |

sprice@hsc.vcu.edu Dear Patrick, You know how much I hate being wrong! Steve P |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
R. John Howe |

Date |
: |
11-24-1999 on 08:24 a.m. |

Dear Pat et al - I like your careful logic although I fear that in the interest of clarity we may be confusing less experienced folks. This is probably something best explained by Marla and perhaps she will step in at some point and clarify for us all. My understanding is that the hypothesis about structure limited designs focuses on the likely direction in which a particular design flowed as between flatweaves and pile weavings. I do not think there is any necessary associated claim that no pile weaving could have preceded the oldest flatweaves known. The claim is that since pile weaving is digital and that almost any pattern could be produced with its "dots," if we find a pile weaver making a design in a way that would have been required in flatweave but that is not required in pile, we can likely presume that THIS PARTICLUAR DESIGN flowed from the flatweave version to the pile version. I do think that many who make this argument also often think it likely that flatweaves preceded pile weaving (I think John Wertime has proposed a specific sequence in his article in Hali 100, September 1998, p. 86-97.) but I do not think this broader claim is entailed in narrower versions of the structure-limited design thesis. Regards, R. John Howe |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Yon Bard |

Date |
: |
11-24-1999 on 08:54 a.m. |

I'd like to quibble with some statements made in this thread: 1. 'Pile weaving is digital, hence almost any design can be achieved.' This would be true if the resolution (knot count) were very high. In practice, design flexibility is quite limited. For example, on many a weaving all diagonals are at the same angle (or, with offset knotting, two angles). Besides, flat-weave (actually, any warp and weft based artifact) is also digital, and because flat-weaves don't require two warps per 'pixel' they may even offer greater flexibility than pile. A glance at a good Soumac or at a Senna kilim will demonstrate what can be done in flat-weave. 2. While I cannot offhand quote any sources, I am quite sure that the affinity of slit-tapestry weaveings to diagonal designs was common knowledge well before Marla wrote about it. 3. I think people generally assume that flat-weaves predated pile because the logical succession of technologies appears to go from uniform plain-weave to striped fabrics to slit-weave to supplementary wefts, which when cut into individual tufts become pile. Whether this is correct, who knows? Regards, Yon |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
R. John Howe |

Date |
: |
11-24-1999 on 04:28 p.m. |

Dear folks - This just to acknowledge that Yon is correct in pointing out that we are talking about relative restriction even with pile weaving. The pile weaver is not in the same position as an impressionist painter but must place her/his "pixels" on a grid. This grid does restrict where the pixels can be placed in relation to one another. Regards, R. John Howe |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Daniel+Deschuyteneer |

Date |
: |
11-26-1999 on 01:39 p.m. |

Daniel Deschuyteneer Dear Steve, I would want to correct your descriptions of weft-faced and warp-faced weaves. It's not the thickness or the size of the yarns which determines the type of weave but the flexibility and the predominance of warps over wefts (warp-faced weave) or wefts over warps (weft-faced weave). In a weft-faced weave wefts aren't heavy. It's in fact the warps that are heavy. The wefts are thin because they must be flexible and they outnumber the warps. When they are closely packed down, they cover and hidden the warps. The horizontal patterning is done by the wefts. On the other hand in a warp-faced weave it's not the thickness of the warps which determines the type of weave, but the number and the spacing of the warps stretched on the loom in relation to weft. In a warp faced weave the warps outnumber the wefts. The warps are tough and flexible and are jammed so tightly that the wefts are hidden. The patterning is made by the vertical stripes of the warps. In a balanced plain weave, the yarns are similar in size, number, flexibility and spacing. Cordially, Daniel |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Marvin Amstey |

Date |
: |
11-27-1999 on 01:22 p.m. |

mamstey1@rochester.rr.com I think Yon and John overstate the limitations of a grid needed for design. Look at Chinese pieces from the 17th c. or earlier. They have very curvilinear designs with only 30 kpsi and often use 2 warps and 2 wefts (offset and warp sharing)per "pixel". Regards, Marvin |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Yon Bard |

Date |
: |
11-27-1999 on 06:16 p.m. |

Marvin, it is relatively easy to give the impression of curvature. It is virtually impossible, though, to draw diagonals except at the discrete angles permitted by the grid. Regards, Yon |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Steve Price |

Date |
: |
11-29-1999 on 09:43 a.m. |

sprice@hsc.vcu.edu
Dear Yon, You said, It is virtually impossible, though, to draw
diagonals except at the discrete angles permitted by the grid. Like so
many things that are obvious to some, I don't understand this. In real
rugs I see diagonals at several angles, so the discrete angles
permitted by the grid at least allow more than one diagonal angle to
exist. Do you mean, the number of possible angles is finite? That doesn't
seem terribly limiting. Or, is there a more serious limitation? If so,
would you expand on it a bit? Thanks, Steve Price |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Marla Mallett |

Date |
: |
11-29-1999 on 12:13 p.m. |

marlam@mindspring.com Dear folks, There are quite a number of ways to draw diagonals in knotted pile, so a rigid grid, even strictly adhered to, is not the limitation it may at first appear. On a piece of graph paper I've drawn a few of the most common methods of making diagonals. The first row (A-H) shows how diagonals look with a balanced knot count--the same number of knots horizontally and vertically. In D, E, and F the knots are offset in successive rows--tied on alternate warp pairs. The second row in the drawing shows a weave with twice as many knots vertically as horizontally, as in many Turkmen rugs and bags. Here the possibilities increase, as it is more practical for vertical steps to utilize several knots. In L, M, and N the knots are offset. Of course when any of these diagonals are thickened horizontally, the line appears to be smoother, as in Q or R. This optical effect occurs when the diagonal is at the edge of a solid area too. The diagonal in Z is formed in exactly the same way as that in A. Intermediate angles can be drawn by alternating steps of different sizes as in S, where the steps are made alternately with two and three knots vertically. A more shallow angle would result if the steps were alternately made with one and two knots vertically. In T I've shown another way: steps alternate in width--alternately one and two knots wide. Still another option: the angle can be changed by offsetting successive groups of knots in various ways, as in V, where three-knot groups are offset by half a knot. These last methods have been used only occasionally on the rugs we encounter, and usually then to articulate unusual motifs. Recently I saw a Tekke mafrash with this kind of detail used for the tails of small animals. -------- On your broader topic, John: The general design transfer principles that I've lectured on ad nauseum and that Jon Thompson has dubbed "Marla's Laws 1 and 2" are: 1. Design influence flows most often from restrictive techniques to freer techniques. (Knotted pile, soumak and embroidery allow the most freedom. They are the most eclectic; they can copy almost anything. Among the most common restrictive techniques are brocading, the warp-pattern weaves, and slit tapestry. Part of the restrictions lie in the processes themselves; others are due to structural limitations.) 2. Designs are the most stable in those techniques that are the most restrictive; conversely, they are the most likely to change and evolve in the freer techniques. (To find the archetypal forms of a great many designs we must study brocade, slit-tapestry and warp-pattern design material. Motifs in these have frequently been copied by pile and soumak weavers, then altered, to suit the weavers' fancy.) Marla |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Yon Bard |

Date |
: |
11-29-1999 on 03:04 p.m. |

Steve, yes, I meant only a finite number. In actual fact, on just about all Turkoman weavings ALL diagonals are at the same angle, with an occasional steeper angle when offset knotting is used. In a Heriz that I have, all diagonals are either at basic or half pitch. A much more finely woven Isphahan has no diagonals at all - everything is curvilinear. I think few of Marla's examples would qualify as satisfactory diagonals - they are too jagged. Of course, the finer the weave, the closer these things come to looking like true straight-line diagonals. The same phenomenon is well known in computer graphics, where elaborate 'anti aliasing' techniques are used to mitigate the problem. Regards, Yon |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Marla Mallett |

Date |
: |
11-29-1999 on 03:18 p.m. |

marlam@mindspring.com Two points: In some weavings, Turkoman's included, diagonals are made purposely jagged, for contrast, i.e. steps are made two knots wide and two knots tall. I didn't draw nearly all of the possibilities: There are Chinese saddle covers with diagonals made with six-knot steps! One such is shown in my book on page 36. Marla |

Subject |
: |
Diagonals vs curves |

Author |
: |
Steve Price |

Date |
: |
11-29-1999 on 04:05 p.m. |

Dear Yon, In describing the situation in something very finely
knotted, you say ...no diagonals at all - everything is
curvilinear. Two points: 1. I think "curvilinear" could be described
as a large number of diagonals (the tangents to the curve) as accurately
as calling it "no diagonals at all." 2. Many Turkmen textiles,
particularly Tekke, Salor and Saryk, are as finely knotted as most Isfahan
rugs. But I do see your point, and don't want to make the discussion too
Talmudic. Steve |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Marla Mallett |

Date |
: |
11-29-1999 on 04:56 p.m. |

marlam@mindspring.com A couple more related thoughts: I am currently studying a Turkmen piece in which nearly all of the field is done with offset knotting. Regular knotting is used in just a few places for contrast. The especially odd thing about the way many of the diagonals are made is this: all are offset in half-knot steps for a distance, then change to full-knot steps, giving the illustion of slight curves at the ends of the lines. I've also been looking at a piece in which the same sort of effect has been used at the tops of guls--diagonals made with full, single-knot steps for a while, then two-knot-wide steps. The lines clearly bend, and the angles change. Of course most Turkmen weavers used the simplest method possible of making diagonals throughout their weavings--single-knot steps--producing an internal consistency that's pleasing. Experimental efforts at altering this approach were not always successful, but in the earliest Turkmen pieces I've examined I've seen quite a lot of improvisation. Marla |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Marla Mallett |

Date |
: |
11-29-1999 on 07:00 p.m. |

marlam@mindspring.com I just looked closely at another old Tekke mafrash...and found diagonals slanting at SIX different angles! Most common throughout are smooth diagonals made with one-knot steps, but there are also more SHALLOW diagonals made with two-knot steps. Then there are STEEP, jagged diagonals made with two-knot steps, three-knot steps, four-knot steps, five-knots steps, and six-knot steps! All used effectively. Marla |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Yon Bard |

Date |
: |
11-29-1999 on 07:14 p.m. |

Marla, I sure would like to see a picture of the Tekke with the six diagonal angles! Regards, Yon |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Marla Mallett |

Date |
: |
11-29-1999 on 10:45 p.m. |

marlam@mindspring.com Yon, It's a Tekke mafrash with a compartment gul design. I'm doing analyses for an upcoming publication and unfortunately don't have permission from the owners to share images at this point. The variety occurs in the borders--both horizontal and vertical--and is quite obvious and a bit startling once you realize it's there. The motifs are stretched so that the same number of pattern repeats appear in the short inner vertical borders and the outer long ones. There is no overlapping of knots, just different-sized steps. The freedom with which these simple, repeated motifs are executed give the small piece a little extra vitality. Marla |

Subject |
: |
RE:Structure-limited designs |

Author |
: |
Steve Price |

Date |
: |
11-30-1999 on 10:28 a.m. |

sprice@hsc.vcu.edu
Dear Yon, Here's one that you put onto a Turkotek discussion a few months
back. I see at least 3
diagonal angles to the right as well as counterparts for each to the left.
Strictly speaking, that makes at least 6 angles although the mechanics of
making the right and left versions of the same angle are identical. Steve
Price |

Powered by UltraBoard 2000 <http://www.ub2k.com/>