A damask is its own complement

Hold a damask tablecloth at arm’s length and turn it slowly. The figure appears, disappears and reappears inverted, and the effect is so complete that it is natural to assume two different weaves are at work. There is one, and it is used twice.

The assumption is worth examining before it is dismissed, because it is what almost every description of the cloth implies. If two areas of a single-coloured fabric look that different, something about them must be different — a different weave, a different tension, a different finish applied to the figure. None of those is present. Every difference the eye reports is a difference in the direction the threads on the surface are running, and the arithmetic below is a list of the things that are not different, each of which would have been a perfectly plausible explanation.

A damask is a satin and its own complement. The figure is warp-face satin; the ground is the same satin with every intersection reversed, which is weft-face satin. Nothing else differs — not the yarn, not the sett, not the float length, not the number of interlacings. The whole of the pattern is carried by which of the two thread systems is lying on the surface.

Why the effect is so strong #

The reason a damask flashes as it turns is that a float reflects along its own length. A warp float lying north–south sends light back to an eye placed north or south of it and looks dark from east or west; a weft float lying east–west does the reverse. Two areas of the same yarn, at the same sett, differing only in the direction their floats run, are therefore about as different as a single-coloured cloth can be made to look — and they exchange appearances completely when the cloth is turned through a right angle.

That is the same specular argument satin’s lustre rests on, applied twice on one surface. What is worth noticing is how little else is available. A single-coloured cloth has no colour contrast to work with, no texture contrast if the binding is the same, and no relief. Direction is the only variable, and damask uses the whole of it.

What the arithmetic says #

Every quantity this site measures from a matrix comes out identical for the two areas, and the identities are worth taking one at a time because each rules out a different explanation for the effect.

The longest float is the same. Complementing a matrix turns warp floats into weft floats of exactly the same length. So the figure is not shinier because its floats are longer; both areas carry the same maximum, and at order eight that maximum is seven.

The interlacing count is the same. The number of face changes in the repeat is unchanged by complementing, because every change is still a change. So the figure is not looser or firmer than the ground.

The shaft count is the same, and so is the threading. The distinct columns of a matrix and of its complement are in bijection, and the bijection preserves which ends share a column. Figure and ground are threaded identically, and the figures here assert that rather than observing it.

The face is shared the opposite way round. At order eight the satin shows seven-eighths warp and its complement one-eighth, and the two sum to exactly one. That is the only measured difference, and it is the whole of the effect.

The consequence nobody advertises #

Because the two areas agree on every structural quantity, a damask has no weak half. Figure and ground wear alike, drape alike, shrink alike and can be set alike, and there is no boundary in the cloth across which any mechanical property changes.

That is a genuinely unusual property for a figured fabric and it is worth comparing against the alternatives. A figure made by adding an extra weft — a brocade, a lappet, an embroidery — puts different material in different places, so the figured areas are heavier, stiffer, and often the first to go. A figure made by contrasting two different weaves puts different float lengths in different places, so the areas abrade at different rates. A damask does neither. The pattern is free, structurally speaking, and it is the only figuring method on this site of which that is true.

It also explains something about how damask is used. Table linen is washed hot, mangled, and worn evenly across its whole surface for decades, which is precisely the service in which a cloth with a weak half fails at its pattern edges. Damask does not have pattern edges in that sense.

Damask, satin and brocade are three different words #

The three get used interchangeably in selling and mean quite different things structurally, and the distinctions fall straight out of what has been established.

Satin is a weave: a single construction with a move number and an order, warp-face throughout, and no figure at all. A satin cloth is one weave everywhere.

Damask is a pair of weaves — a satin and its complement — arranged into areas. There is no extra material, no extra thread system, and both areas are the same weave family. It is figured by reversal.

Brocade is a satin or twill ground with an additional weft laid in over the figured areas only. That extra weft is a third thread system, it is present in some places and absent in others, and it is normally carried on the back where it is not needed or cut away afterwards.

The structural consequence is the one this essay has been about. Damask adds nothing and therefore weighs, wears and drapes the same everywhere. Brocade adds material in the figure and therefore does not, and a brocade’s figured areas are stiffer and heavier and the floats on its reverse are a liability the cloth has to be finished to manage.

There is a fourth term worth separating off because it belongs to the next part of this field entirely. A damassé or figured double cloth makes its pattern by exchanging two complete fabrics rather than by exchanging face and back of one — which is a different construction with a different criterion attached, and it is the point at which figuring stops being a question about one matrix.

Where the contrast actually comes from #

If every structural quantity agrees, the contrast has to be measured somewhere else, and the honest place to put it is the face fraction: the proportion of the surface occupied by each thread system.

At order n a warp-face satin shows (n−1)/n warp and its complement shows 1/n. The difference between them is (n−2)/n, which rises with the order: three-fifths at five, three-quarters at eight, five-sixths at twelve. So higher orders give stronger figures, and that is the arithmetic behind a preference the trade states as a matter of quality.

And that trade is the one this site has met before under another name. A longer float gives more lustre and less resistance to abrasion and snagging, so the same order that sharpens the figure weakens the cloth, and the ceiling on damask’s contrast is the float limit rather than anything about pattern.

The back is the design inverted, exactly #

One consequence of complementation is checkable by anybody with a damask napkin and is a good deal more precise than it first appears.

Turning a cloth over complements its matrix: where the warp was on the face it is now underneath. The figure of a damask is a warp-face satin, so its back is a weft-face satin — which is the ground. And the ground’s back is the figure.

So the reverse of a damask is the same design with figure and ground exchanged, and not merely a fainter version of the front. The two faces carry the same amount of contrast, in opposite senses, and a damask cloth is genuinely two-sided in a way that almost nothing else in weaving is. Table linen is laid either way up on purpose for exactly this reason, and a well-made damask has no wrong side.

This is also a check on the arithmetic rather than only a consequence of it. If the figure and ground were different weaves — different orders, different bindings — the two faces would not correspond, and the reverse would show a design that was recognisably related to the front and not identical to it. That the correspondence is exact is the observable form of the claim that the cloth is one weave used twice.

Which orders are available #

The satin orders a damask may be built on are not a free choice, and the constraint is the one this site established at its foundation.

A satin of order n needs a move coprime with n and not equal to 1 or n−1, and there is no such move at six. So the usable orders begin at five, skip six, and continue seven, eight, nine, ten, and upward. The contrast (n−2)/n rises through that sequence and the longest float, n−1, rises with it.

Putting the two together gives the design space of a self-coloured damask in one line: contrast and fragility are the same parameter. There is no order that gives a strong figure and a short float, because both are functions of n alone and they move together. A designer wanting more contrast has exactly one lever and it costs exactly one thing.

That is a sharper statement than the trade’s, which tends to describe higher orders as finer or better. They are more lustrous and more contrasted and less durable, and which of those matters depends on whether the cloth is going on a table or through a mangle.

Why the order stops at eight #

The essay’s design space is one line — contrast and fragility are the same parameter — and the exchange rate between them can be written down, which says where on the line to stop rather than merely that there is a cost.

Contrast is (n − 2)/n and the longest float is n − 1, so differentiating both with respect to the order,

the marginal contrast per unit of added float is 2 ÷ n².

Doubling the order quarters the return. Going from five to eight buys fifteen points of contrast for three picks of float; going from twelve to sixteen buys four points for four picks. The contrast saturates towards one and the float does not saturate at all, so every step up the sequence is a worse bargain than the last, and the worsening is quadratic rather than gradual.

Which the float limit turns into a hard stop #

The rate says the returns fall away; the float limit says where the line ends, and the two meet at a recognisable place.

A float’s length in the cloth is n − 1 times the pick spacing. At an ordinary linen or cotton damask sett of twenty-four picks to the centimetre that spacing is 417 micrometres, so

an eight-end satin floats 2.9 millimetres and a twelve-end satin 4.6.

Set that against the design rule figured weaving works to — a maximum float of about three millimetres on a cloth that will be used, more on a cloth that will only be looked at — and the arithmetic lands exactly on the classical choice. A three-millimetre float limit caps the order at eight. A five-millimetre limit reaches twelve, and twelve-end damasks exist and are the fine, delicate, decorative end of the trade.

So the sequence the essay lists — five, seven, eight, and upward — is not open in practice. It is bounded above by a float rule and the bound falls between eight and twelve at every sett a damask is woven at.

That is a derived reason for a choice the trade states as taste. The classical damask is an eight-end satin because eight is the largest order whose float survives a mangle at the setts table linen is woven at, and the contrast available at eight is three quarters of the maximum with the last quarter unreachable.

And it says which lever moves the bound #

The float length is the order times a spacing, so the bound moves with the sett as well as with the order — which gives a designer a second lever the contrast argument on its own does not show.

A more finely set cloth can carry a higher order at the same float length. At thirty-six picks to the centimetre the spacing is 278 micrometres, so a twelve-end satin floats 3.1 millimetres — the same as an eight-end satin at twenty-four. The fine damasks are fine in two senses at once, and the second is what pays for the first.

That is the trade this whole ladder keeps producing, in one more disguise. Contrast is bought with float, float is bounded by length, and length is bought with sett — so the real currency of a damask’s figure is thread count, and the order is the intermediate variable through which it is spent.

What the loom has to do #

The shaft count of a damask is the shaft count of its satin — eight for an eight-end satin, and the figures assert it — which sounds as though a damask should be weavable on an eight-shaft dobby. It is not, and the reason is the distinction the dobby essay draws.

An end in the figure must weave warp-face and an end in the ground weft-face, and the same end passes through both areas as the design moves up the cloth. So an end’s behaviour is not fixed for the length of the warp, and any machine that groups ends permanently cannot do it — which is exactly what a shaft is.

A shaft loom’s answer is the block damask: divide the width into blocks, give each block its own set of shafts, and accept that the figure is made of rectangles. Two blocks of an eight-end satin need sixteen shafts, three need twenty-four, and the harness grows with the number of blocks while the figure stays rectangular.

A jacquard’s answer is to give every end its own hook, at which point the boundary between figure and ground can be any shape. The cost of a curved outline is one hook per end, and the cost of a rectangular one is one shaft per block, and those two prices are what decide which machine a given cloth is woven on.

What was counted, and how #

The complement is taken cell by cell from the satin the figure is built on, and the satin itself is generated from its move number rather than drawn — so the pair being compared is produced by one rule and one negation, with nothing typed in.

Four identities are then asserted while the figure draws, and they are asserted rather than reported because each of them is a claim the essay makes in prose:

Figure and ground need the same number of shafts. Figure and ground have the same threading, compared position by position rather than by count. Figure and ground carry the same longest float. And their warp-face fractions sum to exactly one.

The last is the one worth having as an equality rather than an inequality. It says the two areas partition the surface, which is the sense in which the pattern is carried by nothing but direction, and it would fail immediately if the complement had been taken of the wrong object — of the threading, say, rather than of the draft.

Where the model stops #

Nothing here says anything about how the figure looks. The matrix decides that the two areas differ in direction and by how much of each system is on the surface. Whether the resulting figure reads as a rose or as a blur is a question about the outline, the scale, the yarn’s own lustre and the light, and this site has no machinery for any of it.

Self-colour is an assumption. Everything above holds for a damask woven in one colour, which is the classical cloth. Weave warp and weft in different colours and the two areas differ in colour as well as direction, and the whole argument about structural identity survives while the argument about where the contrast comes from does not.

A real damask is rarely a pure complement. Large ground areas are often woven in a different satin order from the figure, and outlines are frequently reinforced with a stitching weave to sharpen the edge. Each of those breaks one of the identities above deliberately, and the cloth is the better for it. The pure case is the one the arithmetic is about — and at its edges the pure case is not the best one either: the float that crosses the line between figure and ground is one pick longer than anything inside either satin when the ground is the exact complement, and no longer when the ground starts at any of the other picks it could.

Who found it, and when #

The cloth is much older than any account of it. Damask is named for Damascus and reached Europe through the mediaeval silk trade; the drawloom techniques for weaving it are older still and Chinese in origin, and self-patterned monochrome silks predate the name by centuries.

The description in this essay — figure and ground as a matrix and its complement — is a modern reading, and it is the reading that makes the cloth’s properties fall out rather than needing to be listed. It also makes the historical sequence make sense: the drawloom came first because a curved outline needs individual control of ends and nothing else will do, and every shaft-loom damask ever since has been an attempt to approximate a curve with rectangles at a price a smaller shop can pay.

Where the ladder goes next #

This ladder has followed the loom from the harness to the hook. The rest of the field leaves the loom entirely and goes after the constructions the matrix itself cannot express — a pile is a third thread system, and a leno has no fixed column order at all — which is where this site’s central encoding reaches its boundary rather than its budget.