Field

Compound and figured cloths

Double cloth, pile, leno, and the loom that decides what can exist. Where a fabric stops being one binary matrix — a third thread system, a warp order that changes between picks — and where the integrity criterion runs out and friction takes over.

A 2/2 threading, run out to three widths. The same four shafts threaded for repeats of increasing width. The threading line runs up and back down; the harness never grows, and the number of ends it carries is bounded by the loom's width rather than by its shafts.

The harness does not grow

Nearly every account of a loom says the shaft count limits the repeat. It limits the repeat of a straight draw, and of nothing else — a reversed twill on ninety-six ends weaves on the same four shafts as the twill it was made from, and the ratio grows without bound.

The draft for herringbone, as a loom holds it. The herringbone written the way a weaver writes it: the threading above, saying which shaft each end hangs on; the lifting plan to the right, saying which shafts rise on each pick; and the cloth below, which is not copied from the weave but produced by running those two against one another and then checked against it.

What a dobby stores

A pattern chain does not store picks. It stores lifts — the distinct sets of shafts a draft ever raises — and for most drafts worth weaving that number is very much smaller than the number of picks, which is the second half of the reason a wide repeat is affordable.

The draft for 8-end satin, as a loom holds it. The 8-end satin written the way a weaver writes it: the threading above, saying which shaft each end hangs on; the lifting plan to the right, saying which shafts rise on each pick; and the cloth below, which is not copied from the weave but produced by running those two against one another and then checked against it.

A jacquard is every end its own shaft

The machine is usually described as the one that weaves anything. What it actually does is abolish one of the loom's two budgets and make the other proportional to the width of the repeat — which turns a constraint on the design's kind into a constraint on its size, and those are very different things to be short of.

Figure and ground on 8 ends. A damask is a satin and its own complement. The figure is warp-face and the ground is weft-face, they carry the same longest float and need the same shafts on the same threading, and the whole of the pattern is carried by which system is on top.

A damask is its own complement

The pattern is carried by direction alone. Figure and ground are the same satin, one warp-face and one weft-face, with the same longest float, the same shaft count and the same threading — so the cloth's most famous effect costs it no structural difference whatever between the two areas.

A V-fastened tuft. A cut pile bound into its ground by V fastening, drawn in section. The pile end passes beneath 3 of the 6 ground picks and wraps 1 half-turn around them in all. The integrity criterion says the tuft is attached; how hard it is held is a different question with a different model behind it.

Pile is a third thread system

Velvet, corduroy, plush and carpet all carry threads that are not woven into the ground in the ordinary sense. Two of the three systems make a perfectly good matrix and the third is not in it — so the encoding this whole site rests on has nothing to say about the part of the fabric a hand actually touches.

What a tuft's wrap is worth at μ = 0.3. The two classical ways of binding a cut pile into its ground, with the holding force each provides as a multiple of the tension applied to the free end. The model is the capstan equation and the friction coefficient is measured rather than derived, so the ratio between the two matters and neither absolute number should be quoted alone.

How a tuft is held

A V-fastened tuft and a W-fastened tuft are both attached, and the integrity criterion returns the same verdict for both. One of them is specified for hotel corridors and the other sheds. What separates them is not a topological property at all — it is an angle, and it needs a coefficient somebody has to measure.

Corduroy: floats of 6 ends, cut. 3 wales of an extra weft floating over 6 warp ends each, cut at their midpoints so that each half stands away from the ground. The pile height is half a float's length and nothing else decides it, so a finer sett at the same float count gives a shorter pile; the wale spacing is the float plus its binding ends.

Corduroy is a cut float

An extra weft floats over some ends, a knife runs along the cloth, and each half of the float stands up. So the pile height is exactly half the float's length — which makes corduroy the one pile fabric whose surface is decided entirely by a quantity this site has been counting since its first essay.

Terry at a let-off of 5 to one. Towelling has two warp beams. The ground warp is held at ordinary tension; the pile warp is let off 5 times as fast, and the excess has nowhere to go but up. The loop height follows from the ratio and the pick spacing, and no float length appears in it.

Terry needs two beams

A towel's pile is not woven longer. It is fed longer — from a second warp beam let off several times faster than the first — so the loop height follows from a ratio and a pick spacing, and the float length that decides a corduroy does not appear in it anywhere.

One cloth, before the knife. Double plush: two ground cloths woven face to face with a pile warp shuttling between them. Uncut, the pile is the only thing either ground touches the other through, and the integrity criterion says one cloth. Cut down the middle, every pile end is severed and the same criterion says two. The whole manufacture is the deliberate destruction of an integrity the criterion otherwise exists to confirm.

Velvet is cut apart

Two ground cloths are woven face to face with a pile warp shuttling between them, and while it does the whole thing is one cloth — the criterion says so, and the pile is the only reason it is true. Then a knife runs down the middle and the same criterion says two. The manufacture is the deliberate destruction of cloth integrity.

A leno and the open plain weave it is not. Two warp ends and the picks they hold. On the left the doup end passes under its partner between picks and comes up the other side; on the right it never crosses, which is an open plain weave at the same sett. The layer count under each panel is computed by the same criterion that decides every other draft, and it returns the same verdict for both.

Leno is not a matrix

A doup end crosses under its partner between picks and comes up on the other side, so the end at position three on pick two is a different end on pick three. The entry W[i][j] does not name anything. The encoding is not wrong about leno; it is undefined for it.

What holds a pick in

A plain weave grips its weft by the crimp, and the crimp's wrap angle falls towards nothing as the cloth opens out. A leno crossing is half a turn whatever the sett. So the comparison has an exact answer that does not depend on the friction coefficient at all — and no plain weave can ever reach a leno's grip.

The criterion cannot see friction

This site's central check is exact, decidable in linear time, and structurally incapable of distinguishing a carpet from a fabric that sheds. That is not a defect to be repaired — a separation exists or it does not, and there is no margin in it — and the fancy weaves are precisely the constructions that live in the gap.

Heddles per shaft: stripe. The threading of a narrow satin stripe on a broad plain ground, over a warp of 1,200 ends. Each bar is one shaft and its length is the heddles on it. The draft needs 10 shafts however they are loaded; the heaviest carries 500 and the lightest 25, a factor of 20.0. Spending 20 shafts instead brings the heaviest down to 100.

Where the heddles go

A draft says how many shafts it needs and says nothing about how the ends divide between them. On a satin stripe over a plain ground the two ground shafts carry twenty times what the stripe shafts do — and the only cure is to spend shafts, which turns the threading into an allocation problem with an exact answer.

A tube. A two-layer fabric 6 ends wide in each layer, with the weft's path drawn across the section below the draft. 1 piece of cloth, 0 free selvedges, 1 shuttle, and a developed width of 12 ends however the edges are joined. The connectivity of the infinite repeat is 2 for this draft and cannot tell this construction from the others.

A tube and two cloths are the same draft

Two separate fabrics, a cloth twice the width of the loom and a seamless tube are the same draft to the last square. The check this site is built on reports two layers for all three and is right every time — because the thing that separates them is four free selvedges, or two, or none, and a repeat has no selvedges in it.

Where a stitch can hide

One reversed intersection turns two cloths into one, and half the intersections in the repeat would do it. Almost none of them may be used — a plain-faced double cloth has nowhere at all to put a stitch, a five-end satin has fifteen places or none depending on which rule is asked, and two satins of the same order differ by a factor of two.

The harness a strain budget buys. How many shafts stay inside a 1.0 per cent warp-strain budget, against the clear shed opening the loom needs at the reed: 12 mm gives 43, 16 mm gives 36, 20 mm gives 28, 24 mm gives 21, 30 mm gives 13, 36 mm gives 7, 44 mm gives 1. The shed's tangent enters the strain squared, so the opening is much the strongest thing a loom builder controls.

The harness has a depth

Why does a dobby carry sixteen or twenty-four shafts rather than two hundred? The usual answers are about the mechanism — how many jacks a box can drive, how many hooks a dobby has — and they are real limits that are not the binding one. A stated tolerance on warp strain is a stated distance from the fell, and a stated distance is a whole number of shafts. One per cent buys thirteen.

What holds a tuft in, in newtons

The pile ladder computed a tuft's anchorage as a capstan ratio and said, correctly, that a ratio was all it could offer. A ratio multiplies a tension and there was no tension anywhere on this site. There is one now — and the answer, put beside what a carpet is actually specified at, falls short by a factor of three.

How much too thick a round section is, and what reconciles it. For each cloth in the standard cloth table: the thickness a circular Peirce section predicts, the thickness a cloth of that construction measures, and the force per crossing that makes the flattened model reproduce the measurement. The over-prediction runs from 36% to 81%. The reconciling forces span a factor of 4.6 across a table whose counts span a factor of six, and every one of them is of the order of the contact force the cloth's own warp tension supplies — which is what makes this a model rather than eight fitted parameters. What the rows cannot show is that the thicknesses are trade figures for cloths of these constructions rather than measurements of these particular fabrics, so what is being read is an ordering.

The criterion gets a force

This site's integrity criterion decides exactly whether a draft describes one cloth, and has one standing limitation: it says a tuft bound under one pick and a tuft bound under three are both attached, and it is right, and one of those is a carpet while the other sheds. What separates them needs a normal force in a fabric that is not under tension — the number the rung that computed it recorded as unavailable, and the one a thickness gauge now supplies.

A wetting supplies the force the criterion needs

This collection's criterion decides whether a cloth is one cloth, exactly, and cannot see friction — so pricing what it misses needed a contact force, and the only one available had a measured fabric thickness inside it. A wetted close cloth generates one from geometry alone, and it lands within seven per cent of the measured route's answer.

$A 5.0 mm cotton pile standing and crushed. A pile tuft standing 5.0 mm proud and the same tuft pressed flat, which turns it through a right angle over its own length and so bends it to a radius of 3.18 mm. Its fibres are strained 0.187% — an order of magnitude below the smallest strain anybody has measured a recovery at, so this file declines to say what fraction comes back. The consequence is that a crushed carpet is not held down by its fibres: what keeps a pile flat is the tufts leaning on one another and the friction where they touch. Below 0.469 mm the fibre does enter its measured range, which is the difference between a carpet and a velvet and has nothing to do with what either is made of. What the drawing cannot show is the neighbouring tufts, which are the mechanism.$

A crushed pile is not held down by its fibres

A carpet flattened under a foot has bent its tufts through a right angle, which sounds severe and is not: the fibres in a five-millimetre pile are strained under two tenths of a per cent, an order of magnitude below the smallest strain anybody has measured a recovery at. What holds a pile down is the tufts leaning on one another.

A leno's hole cannot drift

A filter cloth is rated by its largest hole, and the largest hole grows one micrometre for every micrometre an end moves sideways. What holds an end in place in an ordinary weave is friction at its crossings, and that friction falls smoothly to nothing as a cloth opens — with no threshold to warn anybody. A leno's crossing does not: its ends are wrapped through half a turn by construction, so the grip has no sett in it, and at an open cloth it holds eighteen times what a plain weave manages.

Three places a mistake can be made, and three shapes it leaves. A loom holds a design in three separate objects, and a single mistake in each of them produces a fault of a completely different size — not because the mistakes differ, but because of how many intersections each object controls. On a 50 m piece of muslin 1500 mm wide, holding 396,000,000 intersections: one end drawn on the wrong shaft is wrong at every pick for the whole length, 110,000 of them; one pick made in the wrong shed is wrong across the whole width once, 3,600; and one shaft tied wrongly to one treadle is wrong wherever that shaft's ends meet that treadle's picks, which is everywhere — 6,187,500, or 1.6% of the cloth. The ratio between the extremes is 1719 to one, and the largest of the three is the one nobody sees happen, because every thread is exactly where it should be.

Three mistakes and the shape each one leaves

A loom holds a design in three separate objects, and one error in each of them is the same size of error. What they cost differs by a factor of seventeen hundred — and the largest of the three is the one nobody can see happening, because every thread is exactly where it should be.

The pressure a 30° twist puts on its own fibres. A twisted yarn squeezes itself. Every fibre under tension at a radius pulls inward with sin²θ of its tension per unit length, and integrating outward to the surface — where the pressure is zero by definition, because there is nothing outside to push against — gives p(r) = ½σ(cos²θ(r) − cos²α) in closed form. At a surface angle of 30° the pressure on the axis is 12.5 per cent of the core fibre's own axial stress and the mean over the section is 5.65 per cent of it. The shading is that closed form and the curve is the same function plotted; the shape is what matters, and its one uncompromising feature is the zero at the edge. The outermost fibres are held by nothing, which is why a yarn is hairy, why a surface fibre is the one that comes away on a finger, and why singeing changes a yarn's behaviour out of all proportion to the mass it removes.

A tuft is set so it cannot untwist

A cut pile tuft has a free end, and at a free end the pressure holding the twist is zero. So the twist runs out over a computable length, the tuft opens, and a carpet loses its appearance long before it loses any material.

A chenille is a yarn that is already a fabric

A chenille yarn is a woven gauze cut into strips. Every structural question this collection asks about cloth can be asked about it — does it hang together, how many systems has it, what holds the pile in — and the answers are the same answers one level down.

Three kinds of surface, and only one of them starts open. The bearing curves of a 2/2 twill in sheeting, of a terry loop pile, and of a cut corduroy pile of 30 tex, over the first six per cent of a cloth's thickness. The woven curve opens as the square root of the depth and the loop pile's does the same, because a loop's top is a curved thread like any other. The cut pile does not open: it is a flat line at 5.2% of the plan, from a depth of nothing, because a blade severed every tuft in one pass and left every end in one plane. That is not a larger contact area, it is a different kind of contact area — one that does not vanish as the load goes to zero, which is a property no woven surface has.

A pile is the only surface with no crowns

Every other fabric touches on the tops of curved threads, so its contact vanishes as the load goes to zero and the pressure on what is touching rises without limit. A cut pile touches on flat ends left in one plane by a blade — so its contact is finite at no load at all, its pressure concentration is bounded, and it is the only fabric whose abrasion mass loss is an honest measure of its damage.

Hair, nap and pile are one construction

There are three surfaces made of fibre standing off a cloth, and they have been arrived at from three different directions and in three different fields. They are the same object at three settings, and what separates them is not what they are but how much of them anybody decided.

The fold the trade actually makes. 2 singles of 20 tex cotton at 800 turns a metre, folded at 566 — a ratio of 0.707. That is 1/√2, the ratio at which the fold's surface helix angle equals its singles' — 22.8° — because a fold of 2 singles is √2 times the diameter. The trade's own bracket for 2 folds is 0.6 to 0.75, and it contains this number.

A leno twists what a weave only crosses

Every woven cloth's threads have a linking number of zero, and that is why an open cloth slips. A leno is the one woven structure whose warp ends wind about one another, so it is the one whose threads are linked — and it is famously the structure that holds at setts where nothing else does.

A braid is a third way to hold threads

Weaving holds by friction and knitting holds by linking. A braid does neither: its strands travel across the structure and back, so no pair of them is linked and no pair of them returns to where it started — and it holds without a reed, a beat-up or a sett.

A two-layer interchange, 1 block by 2. A two-layer cloth whose layers change places from block to block, drawn as the draft and as the section a weaver would draw. Each block is one repeat of the stack, 4 picks by 4 ends, and the design has 2 boundaries in it. Every strand in the draft is coloured by the cloth this site's criterion puts it in, and there is one colour, because there is one cloth — with no intersection reversed anywhere in the repeat. A stitch joins two layers at a point; an interchange joins them along a line, and the line is the design's own block boundary rather than anything added to it. The section below is schematic: it draws each ply as a line rather than as its threads, because what has to be seen is that the two lines cross, and a section at thread level over 2 blocks is a picture nobody can count.

An interchange joins what a stitch would have had to

The rung below spent a whole essay on where a stitch may be put in a double cloth, and found face weaves with nowhere to put one at all. A design in which the two layers change places from block to block needs no stitch anywhere: the boundary is the join. Two blocks side by side make one cloth with not a single intersection reversed, and the same two blocks with no boundary between them make two.

plain over 5-end satin: what each layer's warp does. A warp end of each layer of a double cloth, in section over the same span of cloth, with the length of warp each eats drawn beneath. The crimps come from Peirce's geometry asked about a thread that bends at its own average rate rather than at every crossing, and they are 14.35% for the plain and 2.03% for the 5-end satin. They are not the same, so the two layers consume warp at different rates: a hundred metres of cloth takes 114.4 m of warp from one and 102.0 m from the other. A beam delivers one rate, so the difference has nowhere to go and accumulates with the length woven — one pick spacing after 4 mm. What the sections cannot show is the yarn's thickness, which is exaggerated so the path is legible; the crimps beside them are computed at the real diameter and are not read off these drawings.

Two layers need two beams

A layer weaving a metre of cloth eats one plus its crimp metres of warp, and a beam delivers one rate. A plain face over a five-end satin back differs by twelve percentage points of crimp, which is twelve metres of warp over a hundred-metre piece and one pick spacing of slack after four millimetres of weaving. The difference has nowhere to go and does not settle — so the only double cloth that can share a beam is two layers of the same weave at the same sett.

How many layers the repeat, the harness and the beams each allow. Three ceilings on the number of layers a double cloth can have, for five layer weaves. The repeat's bound is half its ends and is a property of the notation. The harness's is the strain budget — 13 shafts on an ordinary broad loom at a 1.0% warp strain limit — divided by the shafts one layer of that weave costs. The beams' is how many warps the loom carries, which is 2. The shortest of each three is marked, and it is the beams at the coarse end of the table and the harness at the fine end; the repeat is never the binding one except at two-end layers, where it happens to coincide with the harness. A double cloth of eight-end satin layers needs 16 shafts and the budget is 13, so it is a jacquard construction by arithmetic rather than by choice. What the bars cannot show is the pick rate: a k-layer cloth needs k times the picks per centimetre of finished cloth and takes k times as long to weave, which is a cost rather than a ceiling and is the reason four-layer cloths are rare even where they are possible.

The repeat allows four layers and the loom allows two

A repeat of eight ends can hold four separable cloths, and there is a witness that reaches the bound exactly. No loom weaves four. The harness's strain budget buys thirteen shafts and a layer costs its own weave's shaft count, so five-end satin layers stop at two and eight-end satin layers cannot be doubled on a dobby at all — and two differing layers already want a beam each. The notation's ceiling is the only one of the three that is never binding.

A designed thin place does not get worse and an accidental one does. The thinnest place a yarn reaches, against how many gauge lengths of it are tested, for a slub yarn and for a randomly uneven yarn of the same 36.3% coefficient of variation. The slub's floor is its base count — 89% of its mean — and it is a horizontal line, because a designed variation has a stated minimum and never goes below it. The random yarn's minimum is an order statistic and falls without limit: 56% over 10 lengths and 22% over 30000. So the advantage is not a number but a function of how much yarn is being asked about, running from 1.6× to 4.0×. What the curve cannot show is the break itself: a yarn's strength at a thin place is not proportional to its linear density there, and the conversion needs a fibre model.

A designed thin place is kinder than an accidental one

A slub yarn and a badly spun one can carry exactly the same coefficient of variation, and the number tells a mill nothing about which it has. The designed variation has a floor — its base count, and it never goes below it — while the accidental one has a tail that falls further the more yarn is tested. At 36% CV the slub bottoms at 89% of its mean and the random yarn reaches 24%, and the gap widens from 1.6 to 3.7 times as the test grows from ten gauge lengths to ten thousand.

A bouclé at an overfeed of 0.80. A core of 20 tex running straight, an effect thread of 30 tex delivered 80 per cent faster than it, and a binder of 15 tex over the top. The surplus effect thread has nowhere to go but into loops, and their size is not a free choice: a semicircular loop of radius r consumes r(π − 2) of surplus, so at one loop every 3 mm the radius is 2.10 mm — 0.70 of the spacing. The yarn's resultant count is 89 tex and the sum of its three components is 65, because the effect enters multiplied by its own overfeed. What the drawing cannot show is the load path: the core and the binder are at the yarn's own length and the effect thread is longer than the yarn, so a pull on the yarn is carried by 39 per cent of its mass and the loops stay slack.

A fancy yarn has its crimp in the wrong thread

A bouclé is made by feeding an effect thread eighty per cent faster than the core it wraps, so the yarn contains more thread than it is long — which is crimp by this collection's own definition. A cloth's crimp is in the thread that carries the load, so removing it is the cloth's first extension. A bouclé's surplus is in a thread that carries nothing, so pulling the yarn stretches the core at once and the loops never straighten. Thirty-nine per cent of the yarn's mass is on the load path, and the other sixty-one is decoration.

An 8-end weft-faced satin through a point tie on an end. The 8-end satin on a move of 3, weft-faced, woven through a point tie of 12 hooks that turns on a single end, so 22 ends carry one repeat. The strip above is the tie, hook number against end. Inside the satin the weft floats 7 ends; across the turn it floats 13, outlined, because on some pick the nearest interlacing is 7 ends away on both sides. What the drawing cannot show is how a long float at one line down the cloth behaves in use, which depends on the yarn and the finish rather than on the count.

A point tie nearly doubles the float at the turn

A point tie halves the hooks a symmetrical figure needs by driving every hook's end twice, out and back. It mirrors the ground as well, and a satin is symmetrical about no end: across the turn its weft floats 2n − 3 ends, or 2n − 2 between ends, at every hook count and wherever the satin starts. At eight ends that is thirteen or fourteen against seven — and the turn's other promise, more repeats that fit the width, is not kept either.

A lancé weft bound where the 8-end satin interlaces. A warp-faced 8-end satin on a move of 3, with an extra weft thrown after every ground pick and laid on the face over figures 12 ends wide and 28 ends apart. Between figures it is bound only on ends the ground already drops on an adjacent pick, 56 places in the drawn repeat, which leaves back floats of at most 4 and no binding on the face. The drawing is 40 ends by 16 picks and counts as 1 cloth. What it cannot show is whether a binding point counted as hidden is hidden in a real cloth, which depends on the yarns' contrast and on how closely the ground's floats cover it.

A brocade weft floats as far as the next figure

A brocade's pattern weft is a third thread, laid on the face only where the figure wants it. Thrown from selvedge to selvedge it floats on the back across the whole gap to the next figure, unless an end is dropped under it — and on a satin ground a dropped end hides only where the ground already interlaces. That fixes the shortest float that shows nothing, at the larger of the move and its complement less one, and puts the smoothest satin ground at odds with the best-bound brocade.

A jacquard harness 130 cm wide under a 150 cm fall. A jacquard harness in front elevation, to scale: hooks spread over 40 cm, cords fanning down 150 cm to a comber board 130 cm wide, and hanging straight from the board to their mails. The centre cord is vertical and lifts its mail by the whole 10 cm hook lift; the edge cord leans 16.7° and lifts its mail 9.60 cm, 96.0% as far, and its bend at the board adds 9% to the hook's load at a friction coefficient of 0.3. What the drawing cannot show is the tie that decides which hook feeds which hole, which in a real mount is not the simple spread assumed here.

A jacquard harness needs three half-spans of height

A jacquard has no front shaft and no back one, so every end takes the same shed — as long as every cord hangs straight. Across the width they cannot: the hooks sit in a machine a few tens of centimetres wide and the comber board is as wide as the cloth, so an edge cord leans and its mail rises by the difference of two hypotenuses rather than by the hook's lift. Holding the edge shed within five per cent of the centre's takes a fall of about three times the edge cord's sideways reach, and that is a height a room has to have.

$20 tex yarn in one layer and in 2. Sections across the width, to scale, of a cloth of 20 tex cotton at a cover of 0.8, and of the same yarn per area divided into 2 layers two ways: by count, 10.0 tex at the same sett, and by sett, 20 tex at 1/2 of the ends. Divided by count the cloth is 1.41 times as thick with a cover of 0.57 in each layer; divided by sett it is 2.00 times as thick with a cover of 0.40. At the free end of the yarn's stiffness bracket both are exactly as stiff as the single cloth; at the coherent end the first is 0.50 times as stiff and the second 1.00. What the sections cannot show is crimp, which thickens every layer by an amount the weave decides.$

A double cloth is only softer if its yarn is set

A double cloth is sold as weight without stiffness: two light cloths in place of one heavy one. Divide the same yarn into two layers and the cloth is √2 or twice as thick, but at the free end of a yarn's stiffness bracket — where an unset yarn sits — its bending rigidity does not move at all, because it is the number of fibres across the width times the stiffness of one. Only a set yarn makes the double cloth the softer, and stitching the layers together pushes it the other way.

The doup end pays for the crossing

A leno holds its picks because its doup end crosses under its partner and comes up on the other side — half a turn at every crossing, whatever the sett. That half turn is also a length. When 0.25 mm ends cross every millimetre, the doup end travels 22.5 per cent further than its partner: twenty-two metres of extra warp over a hundred-metre piece, and a shared beam a pick short within four and a half millimetres of cloth. The extra falls with the square of the crossing interval, which is why a leno is woven from two beams and its crossings are spaced as far apart as the cloth allows.

A bouclé cloth jammed at its loops and at its count, overfeed 0.8. Sections across 36 mm of plain-woven cloth in a bouclé of 89 tex — a 20 tex core, a 30 tex effect overfed 80% with loops every 3 mm, and a 15 tex binder — drawn to scale. The loops stand 2.10 mm off the core, so the yarn's outline is 4.43 mm across while the diameter its count implies is 0.352 mm, 12.6 times smaller. Jammed at the outline the cloth takes 1.13 ends a centimetre and weighs 20 g/m²; jammed at the count, 14.2 and 252 g/m². What the drawing cannot show is where between the two a real cloth jams, which is how far its loops interleave with their neighbours'.

A bouclé is set by its loops and weighed by its count

A bouclé yarn has two diameters: the one its count implies, a third of a millimetre, and the one its loops occupy, over four millimetres. A cloth jams where its yarns touch, and a bouclé's yarns can touch at either — so its jamming sett is a bracket twelve and a half times wide, from 1.1 to 14 ends a centimetre. The weight follows the count wherever in that bracket the cloth is set, which puts the open end at a twelfth of the close end's weight with four per cent of its area covered by yarn.

The crossed shed, with the back standard 4 shafts behind the doup. A leno's crossing end in section from the fell to the back rest of an ordinary broad loom, in the crossed shed. The doup 300 mm from the fell lifts it 50.0 mm; its back standard 64 mm further back leaves it 60.7 mm down, beneath its partner. Between the two eyes the end climbs 110.7 mm in 64 mm, and the whole path is 70.2 mm longer than the straight line, 5.85% of the free warp, of which 63.8 mm is that one climb. The same end in the open shed, lifted at its back standard, is 7.2 mm longer, and an ordinary end on the doup's shaft 5.5 mm: the crossed shed is 12.7 times the ordinary shed. Vertical scale exaggerated 3 times. What the drawing cannot show is friction at the eyes, which would keep the extra length in the short span between them.

An easer gives back the kink the crossed shed puts in

A leno loom gives its crossing ends slack at every crossed shed, and the obvious reason — the doup carrying its end sideways — is worth a hundredth of an ordinary shed. The length is one heddle further back. In the crossed shed the doup lifts the end on the far side of its partner while the end's own back heddle, a few shafts behind, holds it down, and between those two eyes the end climbs eleven centimetres in six and a half. On a loom that stretches an ordinary end 0.46 per cent, that is 5.85 per cent — nearly thirteen times as much — and no position of the back heddle takes it below four.

Bouclé loops on 3 mm binder spacing at overfeeds of 30%, 80%, 150%, 300%, 500%. Loops of an overfed effect thread drawn to one scale, each between two binder points on a straight core, as the curve of least bending energy with its ends along the core. Overfeed 30%: 1.30 times its base in thread, 1.10 mm tall, a bell, against a semicircle of 0.79 mm; Overfeed 80%: 1.80 times its base in thread, 1.95 mm tall, a bell, against a semicircle of 2.10 mm; Overfeed 150%: 2.50 times its base in thread, 2.92 mm tall, overhanging, against a semicircle of 3.94 mm; Overfeed 300%: 4.00 times its base in thread, 4.83 mm tall, overhanging, against a semicircle of 7.88 mm; Overfeed 500%: 6.00 times its base in thread, 7.25 mm tall, overhanging, against a semicircle of 13.14 mm. The dashed arcs are the semicircles, drawn where they fit. What the drawing cannot show is the binder's own path and thickness, which the loop's feet are taken to sit exactly on.

A bouclé loop is an elastica, not a semicircle

A bouclé's loop was taken to be a semicircle, and the semicircle cannot be right: it meets the core at a right angle, and at eighty per cent overfeed it is wider than the gap it stands in. Solved as what it is — a length of thread leaving the core along the core at two binder points — the loop is taller than the semicircle below an overfeed of two thirds and shorter above it, overhangs at 119 per cent and closes its neck at 559. Nothing of the fibre is in its shape, and the bouclé's jamming bracket at eighty per cent is 11.7 wide rather than 12.6.

A bouclé loop at 80 per cent overfeed, pressed to four heights. One loop of a bouclé at 80 per cent overfeed on 3 millimetre binder spacing, drawn to scale at 100, 80, 50, 20 per cent of its free height, with the force it pushes back with beneath each. The pressed loop is the same elastica the free one is, with one condition added: at its apex the tangent is along the core again, so a pressed loop is two half-loops each with its rise prescribed. At 100 per cent it stands 1.95 millimetres and pushes with -0.00 millinewtons; At 80 per cent it stands 1.56 millimetres and pushes with 12.19 millinewtons; At 50 per cent it stands 0.97 millimetres and pushes with 10.02 millinewtons; At 20 per cent it stands 0.39 millimetres and pushes with 4.90 millinewtons. What the drawing cannot show is the stiffness bracket: the force is the lower bound, with the fibres free to slide.

A loop has a maximum force in it

A bouclé loop pressed by its neighbour is the same elastica the free one is, with one condition added — at its apex the tangent lies along the core again, so a pressed loop is two half-loops with their rise prescribed and needs no contact solve at all. Solved, the eighty-per-cent loop pushes back with nothing at its free height, twelve millinewtons at four fifths of it, and two and a half at a tenth. A curve with a maximum in it is a softening spring, and the promise this account made — a cloth's position read from its beat-up — cannot be kept, because the whole bracket sits at one pressure.

A 100-gram bouclé in section, with the depth to its own load path. A bouclé cloth of 100 grams a square metre in section: the core and binder along the bottom, carrying everything, with the effect thread's loops standing 0.31 millimetres above them. The cloth's surface is the loops and nothing else — its loop envelope is 12.6 times its count diameter — so a rubbing surface meets effect thread first and reaches the core only after that depth. 60.7 per cent of the yarn's mass is effect thread on no load path and 39.3 per cent is core and binder. What the section cannot show is the third dimension: the loops of neighbouring yarns lie between these and are pressed by them.

A bouclé wears from the loops down

A cloth loses its strength before its mass, because a woven cloth's crowns are the very threads that carry the tension. A bouclé inverts it exactly. Its surface is its loops, its loops hang off a core that carries everything, and 61 per cent of its yarn is on no load path at all — so a rub takes three tenths of a millimetre of thread that does nothing before it reaches anything that does, and the cloth is ruined to look at while it is still as strong as it was woven.

A comber board for 60 ends a centimetre, in side elevation. A jacquard's comber board seen from the side, with the fell of the cloth at the left and the back rest at the right. The board carries one hole per end; at 60 ends a centimetre the ends are 0.167 millimetres apart and a cord with a mail on it needs 0.9, so the holes are ruled in 6 rows staggered fore and aft, 6 millimetres apart — a harness 30 millimetres deep. Each row's ends are strained by its own distance from the fell: 0.460 per cent at the front and 0.524 at the back, a spread of 0.0636. What the elevation cannot show is the sideways fan of the cords above the board, which is a separate and much larger geometry.

A jacquard's harness has a depth after all

A jacquard was said to escape the shaft loom's depth entirely, because every mail hangs at the same distance from the fell. Every mail does, if the comber board has one row of holes — and it cannot. At sixty ends a centimetre the ends are a sixth of a millimetre apart and a cord with a mail on it wants most of one, so the holes are ruled in six rows thirty millimetres deep. The escape is real and it is a factor of eight rather than a release, and it closes as the cloth is set finer.

A damask is the only figure that costs its beam nothing

Figure and ground consume warp at different rates, and the difference accumulates down the length of the figure. An eight-end satin figure on a plain ground puts its two regions fourteen per cent apart, which on a loom absorbing a millimetre of slack bounds the figure at eight and a half millimetres. A damask's two regions are the same satin used two ways, so complementation leaves both float lengths where they were — and their crimps are identical rather than close, to the last bit of a double.

The crossed shed's three spans at an eye friction of 0.3. A leno's crossing end in the crossed shed on an ordinary broad loom, back standard 4 shafts behind the doup, with both heddle eyes gripping at a coefficient of 0.3 and a background tension of 0.5 N. The end turns 69 degrees at the doup and 64 at the back standard, capstans of 1.44 and 1.40. From fell to doup it settles at 6.95 N, a strain of 5.73%; from doup to back standard it settles at 9.48 N, a strain of 7.97%; from back standard to back rest it settles at 6.78 N, a strain of 5.57%. With frictionless eyes every span would take 7.08 N; an ordinary end at the doup takes 1.02 N and the yarn breaks at 3.74 N on its initial modulus. Vertical scale exaggerated 3 times. What the drawing cannot show is the tension's fall round each eye, which happens over the eye's own few millimetres.

A heddle eye lets the kink through

A leno's crossing end is pulled up at its doup and held down at its back standard, and the length that costs was priced as if both heddle eyes were frictionless. They grip, and gripping ought to trap the kink between them at nearly a hundred per cent strain. It does not come close. A capstan bounds a ratio of tensions, not a difference, and the spans either side are already stretched, so at a coefficient of 0.3 the span between the eyes takes 9.48 newtons against 7.08 with no friction at all — a third more, not fifteen times more — and an easer has to give back 67.9 millimetres rather than 64.4. What friction changes more is when the length is wanted.

The beams a disc 12 blocks across needs, column by column. A disc 12 blocks across, 12 blocks by 12, figure shaded, with the beam feeding each column of ends written above it, for a figure floating 8 on a ground floating 1, whose warp crimps are 14.15 per cent apart, at 4 mm blocks over a 100 m piece with 1.2 mm of slack. It has 4 distinct columns and 4 distinct shares of the repeat in figure, and needs 4 beams. What the grid cannot show is the cloth's own crimp interchange, which would move tension between the columns instead.

A figured warp needs a beam for every share of its figure

Figure and ground take up warp at different rates, so a figured cloth on one beam is bounded in how long its figure may run, and a second beam is the obvious escape. It escapes only for ends that live wholly in one region. An end that crosses the figure for part of the repeat consumes warp at its own rate, and two ends can share a beam only if they spend the same share of the repeat in the figure and never drift a slack apart inside it. So a round figure twelve blocks across needs four beams, ninety-six blocks across needs twenty-nine, and any crimp difference at all — a third of a per cent will do — costs every one of them over a piece.

What each further binder takes off a snag. How far a snag drags thread through a bouclé, in binder points, against how many binders the yarn carries: 1, 7.43 points; 2, 3.72 points; 3, 2.48 points; 4, 1.86 points; 6, 1.24 points. The relation is exactly inverse, because the grip is the exponential of the wrap angle and the reach is a logarithm of it, so the second binder removes 3.72 binder points and the third a further 1.24.

The second binder buys half of everything

A bouclé's loops are held by their binder, and a snag drags thread from one loop to the next until the thread breaks. The essay before it left that reach at seven binder points on one binder and said a second would square the grip. It does — and because the grip is exponential in the wrap and the reach is a logarithm of it, the reach is exactly inverse in the binder count: 7.43 points, then 3.72, then 2.48. Half of everything any number of binders can buy is bought by the second one, which is how many the trade uses.

$What two layers of one yarn buy. For each cloth in the table here, the thickness and the thermal resistance of a two-layer cloth carrying exactly the same yarn per unit area, as ratios to the single cloth. The thickness ratio is √2 everywhere, to three figures. The warmth ratio runs from 1.588 on the openest cloth to 1.717 on the closest, because dividing the yarn also divides the fibre fraction and the mixture conducts less.$

Two layers are warmer than they are thick

Six essays have taken a double cloth apart from the draft's side. None has asked what the reader gets. Divide one cloth's yarn into two layers and the fabric is 41 per cent thicker — √2, which is the account's own law — and about 70 per cent warmer, because dividing the yarn also divides the fibre fraction and the mixture conducts less. The gap widens with every further layer and never closes, and a shaft loom stops at two.

$What the crossing end asks for, against what a cam gives it. The length a leno's crossing end needs at each fraction of the crossed shed's opening, against an easer driven in proportion to the shed. The demand is nothing for the first eighth of the opening and then rises steeply, because the kink's extra length is a hypotenuse less its run and grows as the square of the climb. The proportional easer runs ahead of it everywhere, worst at 0.375 of the way open, where it has given 16.5 millimetres more than the end can use.$

A cam easer gives its slack too early

The essay before it left the crossing end with fifteen millimetres it could not use at half the shed, and three things it might do with them. All three are decidable. It cannot reach the shuttle, because the same eyes that trapped the kink keep the slack behind the harness — and the margin is negative, so that friction is not a nuisance here but the reason a leno weaves at all. It cannot snarl, because the span is under half this account's own threshold. It sags seventy-two millimetres where the easer gives it, and the repair is a cam cut to the demand rather than to the shed.

A stitch takes back what the parallel rule cannot see

Dividing a cloth's yarn into two layers makes it warmer than its thickness, and the essay that found it computed the warmth by the one mixing rule that cannot see a stitch — the one that treats every fibre as though it ran straight through the cloth. That rule gives the largest bonus there is. Yarn in a cloth lies in its plane, and by the rule for fibres lying across the heat's path a double muslin's bonus is 8.4 per cent, not 18.8. A stitch is the one fibre that does run through the cloth, and at half the intersections it takes the whole bonus back.

A blade leaves loose fibre in every tuft

The account of hair, nap and pile ended on a clean claim: a blade collapses a population's length to one value, so a cut pile has no tail and nothing in it reaches past the rest. The tips are one length. The fibres are not. A tuft cut from staple yarn is a length of yarn with fibre ends scattered along it, and every fibre end that lands in a leg leaves a piece between it and the blade that nothing in the draft holds — a fifth of the pieces in an ordinary wool carpet, none in a filament one.

A figured warp pays in tension before it needs a beam

Figure and ground take up warp at different rates, and counted against a fixed slack, any difference at all — a third of a per cent — made a round figure need four beams. A real let-off holds tension, not length. Ends that consume more pull harder, and an end pulled harder gives up crimp, until every end consumes alike. For an eight-end satin figure on a five-end satin ground that costs a hundredth of a newton an end, and one beam serves. For a satin on plain it costs more crimp than the ground has, and no tension will do.

$The springs a leno easer can be. The stiffest spring, per end, that an easer can carry without any span of the crossing end rising above an ordinary end's working tension at any fraction of the crossed shed, against how far its preload sits above the warp's resting tension of 0.5 N. Behind the harness, where an easing bar acts, it is 1.1 N/m at a preload equal to the resting tension and falls to nothing by 15 per cent above it; in front of the harness it is 4.8 N/m. Everything under a curve works and nothing above it does. The window is a corner: a spring barely stiffer than a dead weight, with a preload matched to the warp's own tension. What the chart cannot show is the bar's mass, which the loom's speed makes matter and which the next views take up.$

A leno easer should be a light weight

A spring easer gives length when the crossing end pulls, so it cannot give it too early, which was the fault in a cam driven off the shed. The question left was its rate. The answer is that it hardly has one. Behind the harness the easer feels the back span, and two gripping eyes keep that span within fifteen per cent of its resting tension while the kink carries its load. So the spring must hold its span almost constant over a sixty-eight-millimetre stroke: at most 1.1 newtons a metre per end, a dead weight in all but name. At speed the bar's own mass is what limits it, and it falls with the square of the loom's speed: 2.3 grams an end at a hundred picks a minute, 0.6 at two hundred.

Which loose pieces in a cut tuft are free at once. A cut-pile tuft 12 mm tall on a 2 mm base, spun from a 90 mm staple, with a seeded sample of its fibre pieces drawn along the yarn and the tip's untwisted run-out, 5.6 mm, shaded down each leg. 22 pieces pass under the binding pick; of the 9 that lie wholly in one leg, 3 lie wholly inside the run-out, where the twist presses on nothing, and are free at once; the other 6 reach down into twisted yarn and are held until wear opens the twist further. What the drawing cannot show is the run-out's edge, which is a fall in pressure rather than a line.

A cut pile sheds in two stages

Every cut tuft of staple yarn holds loose fibre pieces that nothing anchors but the twist, and at the tip the twist holds nothing: it runs out over a length the grip arithmetic already gives, five and a half millimetres in a wool carpet yarn. A loose piece lying wholly inside that run-out is free the day the carpet is laid; one reaching below it is held until walking opens the tip's twist. So the reservoir drains in two stages, and the pile height decides the split. A velvet sheds everything it will ever shed at once. A twelve-millimetre wool carpet sheds a fifth of its reservoir at once and four fifths later. A shag sheds almost nothing at first and fourteen per cent of its fibre eventually.

A disc woven on one beam, as it catches a lamp. A disc of 5-float satin on a ground of float 3, twelve blocks square, woven on one beam so that every column is pulled to the tension that makes it consume what the others do, and drawn as its blocks catch a lamp. A block is lit when its warp's crossings still reflect into the eye; outlined blocks are figure. At the mirror every block is lit. Tilted by 10.8° along the warp, the ground in the columns that carry most ground — the tightest — has gone dark, while the ground in line with the figure's widest part, pulled least, still shines: a band as wide as the figure runs down the ground above and below it. What the drawing cannot show is the diffuse light every block also returns, which dilutes the contrast and does not reverse it.

A figure woven on one beam shows its columns off the mirror

A figured warp shares one beam by letting the columns that want more yarn pull tighter until every column consumes alike. With the figure allowed to give crimp as well as the ground, the tightest column of a satin disc on a twill needs 0.013 to 0.042 newtons, under a tenth of the working tension. Its warp lies tens of micrometres lower, a relief too small to see. At the mirror it shines exactly as a slack column does. A few degrees off the mirror, along the warp, its crossings stop reflecting before a slack column's do, and the figure's own columns appear as bands in the ground. Only a damask has none.

$Where the weight on a shared leno easer comes to rest. The tension in the back span of each end of a leno pair — the length between its last heddle and the back rest, where an easing bar acts — at each fraction of the crossed shed, with one bar under both ends weighted at 1.0 N a pair, twice the warp's resting tension, so that it rests on its stop with the shed closed. The two back spans always sum to the weight. Closed, they share it equally. The partner, an ordinary end lying in the bottom of the shed, can take only 11 mm of the bar's travel before it goes slack, which it does at 37.5% of the opening; from there the crossing end's back span carries the whole 1.0 N. On a bar of its own, dashed, the same span stays at 0.50 N. What the chart cannot show is what the crossing end's other spans do with the extra tension, which the next views take up.$

One weight under a leno pair ends on one end

A leno's crossing end wants about seventy millimetres eased to it at every crossed shed; its partner lies in the bottom of the shed as an ordinary end and wants none. One weighted bar under both is set where the closed shed shares its weight between them, but the partner can take only eleven millimetres before it hangs, and it is slack by three eighths of the shed. From there the crossing end carries the whole weight, twice the rest its own bar would hold it at, and its tightest span ends forty per cent over the working tension while the partner hangs thirteen centimetres behind the harness. No weight serves both, because a bar gives every end it carries one length.

A leno partner lifted by its doup's loop. The tension in each span of a leno's partner end at the full crossed shed, against how far the doup's loop lifts it above its own straight line 6 mm in front of its heddle on the doup's shaft. The partner is then three spans: fell to loop, loop to heddle, heddle to back rest. With the loop clear it is an ordinary end, 1.07 N in its front span. At a 4 mm lift the front span carries 1.48 N, +38% on the untouched partner, the span between loop and heddle 1.26 and the back span 1.00; with nothing gripping, dotted, all three would carry 1.12. At 8 mm the front span reaches 1.90 N. What the chart cannot show is how far a real loop lifts its partner, which the doup's standard decides and which is swept here rather than computed.

A leno's slackener eases the partner, not the doup

In a bottom-doup leno the doup's loop passes beneath the partner, and if it lifts the partner on its way up it puts a second turn six millimetres in front of the partner's heddle. A four-millimetre lift bends the partner through 74 degrees in two places, and its front span goes from 1.07 newtons to 1.48 — a tenth of that from the bend and the rest from grip. The back span, which is all an easing bar can feel, does not move. A slackener that drops the doup's standard four millimetres clears the partner exactly, gives the crossing end six per cent of its easing, and costs the shuttle nothing.

A leno's slackener only has to arrive

A slackener that lets the doup down in step with the loop keeps a leno's partner clear at every moment of the shed, and it looked as though a real one would have to track that closely. It does not. A partner's tension only rises as the shed opens, so a slackener can leave its whole drop until three quarters of the way up and the peak does not move at all. Lateness is paid for elsewhere: starting later than 0.58 of the shed rubs the loop more than having no slackener, and above 234 picks a minute a falling weight cannot start as late as the partner would let it.

All essays