After the loom

A twill's floats cannot point a wrinkle

A warp-faced gabardine was expected to keep its horizontal wrinkles even more decidedly than a poplin, because its floats run along the warp. They cannot: a fold counts the turns under its arc, a turn is a turn whichever face it leaves, and every twill and satin presents the same runs to a fold across the warp as to one across the weft. What the weave does instead is lift both of a cloth's radii by a few per cent, and an eight-end satin's by up to a fifth — which narrows the window rather than widening it.

Worth reading first: An unbalanced cloth wrinkles one way and not the other · How sharply a weave lets a cloth fold · A cloth has one budget for two directions.

An unbalanced cloth wrinkles one way and not the other, and it does so for a reason that has nothing to do with its weave. A poplin’s warp runs out of crimp at a fold radius of 4.22 millimetres and its weft at 1.85, because the cloth’s one locus ends sooner in one direction than the other, and every fold of a gentle radius between those two numbers stretches the threads when it lies across the warp and costs nothing when it lies across the weft. Every cloth in that account was a plain weave, so the locus was the only thing that could tell the two directions apart.

The account ended on a prediction for the cloths that are not plain. A warp-faced twill — a gabardine, a drill — lays its ends over several picks at a time, and a fold that lands on a float finds no crimp there to spend. So a fold across the warp of such a cloth should find less crimp locally than the warp’s budget promises, and a fold across the weft about what the weft’s budget promises. In a warp-dense, warp-faced cloth the two asymmetries were expected to point the same way, widening the window; in a weft-faced sateen they might cancel.

Counting the turns says neither happens, and the reason is in the drawing below.

A 2-pick fold across a 3/1 twillA 3/1 twill on point paper with a fold 2 picks wide drawn across it at pick 1. A warp end's crimp is made where it turns from over to under, and the dots mark the turns that fall inside the fold. Averaged over the ends there are 1.00 of them, and 1 of the 4 ends in the repeat have none at all — those ends are marked with a line, and each of them crosses the fold dead straight with no crimp to give up. Over every fold position, 25.0% of end-and-place pairs are like that. What the drawing cannot show is what happens to those ends instead, which is that the whole of the fold's length difference goes into their fibres.Where in the repeat a fold falls, and what it finds therea fold 2 picks wide across a 3/1 twill: the crimp it can spend is made at the turns, and 1 of 4ends have no turn inside itturns in the window, per end: 1.00 on averageends with none at all: 1 of 4over every position: 25.0% of end-and-place pairslongest warp float: 3 picksthe band is the fold; the marked ends have no turn inside it3/1 twill · a 2-pick fold at pick 1
Fig. 1 A 3/1 twill, the weave of a drill, with a fold two picks wide drawn across the warp; the slider moves the fold through the four picks of the repeat. The dots are the turns inside the fold, where an end changes face and where its crimp is made. One end in four crosses the fold with no turn in it at whichever pick the fold is placed — the end whose three-pick float spans it.

A fold counts turns, and a turn has no face

A warp end’s crimp is the extra length it carries for going over and under the picks, and that length is made where the end changes face. Between changes the end lies straight along a float, and a float presses on nothing and carries no crimp either. So what a fold can spend, at the place it falls, is the crimp of the turns under its arc. That was the whole of the count of how sharply a weave lets a cloth fold: slide a window down each end, count the changes of face inside it, and see how often the count is zero.

The count has a property the prediction did not use. A turn is a turn whichever way it goes. An end that comes up from the back to lie over three picks and then goes down again has made two turns, and the window does not ask which face the three picks lay on. A float on the face and a float on the back are both a stretch with no turn in it, and they count alike.

That makes the face of a twill irrelevant to a fold. A 3/1 twill and its reverse, the 1/3 twill on the other side of the same cloth, present a fold with identical runs, because the reverse is the same matrix with every cell’s face exchanged and every change of face is still a change of face. The drill and the drill turned over are, to a fold, one cloth.

The same weave read along the picks

That settles the face. The direction is a separate question: a fold across the weft bends the picks, and it counts the turns along each pick from one end to the next rather than down each end from one pick to the next. It is reading the matrix’s rows where the other fold read its columns, and nothing so far says the rows of a twill look like its columns.

A 2-end fold across the weft of a 3/1 twill. A 3/1 twill on point paper with a fold 2 ends wide drawn down it at end 1, so that the picks bend round the fold. A pick's crimp is made where it changes face from one end to the next, and the dots mark the turns inside the band: 1.00 a pick on average, with 1 of the 4 picks crossing it with none, marked by a line. Over every position 25.0% of pick-and-place pairs are bare, against 25.0% of end-and-place pairs for the same width of fold across the warp. Which face a float lies on is not part of the count, because a turn is a turn whichever way it goes. What the drawing cannot show is how far along a thread friction lets a fold borrow crimp from outside the band.
Fig. 2 The same 3/1 twill folded the other way: a band two ends wide runs down the cloth and the picks bend round it. The dots are the picks’ turns inside the band. One pick in four crosses with no turn — the pick passing under the three ends that float over it — and over every position 25 per cent of pick-and-place pairs are bare, exactly the share of end-and-place pairs bare for the same width of fold across the warp.

They do, and for every regular twill and satin. Each end of a 3/1 twill is up for three picks and down for one; each pick of it is under three ends and over one. The runs along a pick are the runs down an end, read backwards, because a twill’s rows are its columns stepped the other way. The same is true of a satin, where every end is up for one pick and down for the rest and every pick is over one end and under the rest. Read for turns, these weaves are the same in both directions — and since their reverse faces are the same too, a warp-faced twill, its weft-faced reverse and the same twill folded the other way round all present a fold with one set of runs.

So a twill’s floats cannot point a wrinkle. The gabardine’s ends float over two picks, and its picks float under two ends; a fold across the warp finds the first, a fold across the weft finds the second, and they are the same stretch. Whatever makes a gabardine keep its horizontal wrinkles, it is not that its floats lie along the warp.

A weave that does have a direction

The argument needs a weave whose rows and columns differ, and the simplest one is a rib. A 2/2 warp rib lifts each end for two picks and drops it for two, so every end turns at every second pick; every pick, meanwhile, goes over and under alternate ends exactly as in a plain weave. Its columns are a basket’s and its rows are a plain weave’s — the shape a cord that is a stripe with no colour in it makes visible as ridges across the cloth.

A 1-end fold across the weft of a 2/2 warp rib. A 2/2 warp rib on point paper with a fold 1 ends wide drawn down it at end 1, so that the picks bend round the fold. A pick's crimp is made where it changes face from one end to the next, and the dots mark the turns inside the band: 1.00 a pick on average, with 0 of the 4 picks crossing it with none, marked by a line. Over every position 0.0% of pick-and-place pairs are bare, against 50.0% of end-and-place pairs for the same width of fold across the warp. Which face a float lies on is not part of the count, because a turn is a turn whichever way it goes. What the drawing cannot show is how far along a thread friction lets a fold borrow crimp from outside the band.
Fig. 3 A 2/2 warp rib folded across the weft, with a band one end wide. Every pick changes face at every end, so every pick has a turn inside the band wherever it is placed and none is bare. The same one-thread fold across the warp leaves half of all end-and-place pairs bare, because each end runs straight over two picks.

A one-thread fold across the warp of this rib finds no turn on half of its end-and-place pairs; the same fold across the weft finds a turn on every pick, every time. That is a weave with a direction of its own, and a fold across its ribs and a fold along them are different folds.

The rib is not a curiosity. Of the 22,730 four-by-four matrices that hang together as one cloth, 6,418 — 28 per cent — present the same runs both ways, and every named twill and basket that fits a four-thread repeat is among them. The other 72 per cent, ribs and their relatives and most of the irregular weaves nobody has named, read differently down the ends from along the picks. Every three-by-three weave that hangs together, all 102 of them, is the same both ways, so the asymmetry needs at least a four-thread repeat to appear at all. The named weaves happen to be the symmetric minority.

The arc decides, and wrinkles have long arcs

A rib’s direction, and any weave’s unevenness, matter to a fold only if the fold is short enough to see them. The window in the count of a sharp fold was two or three picks wide, the scale of a pressed crease. A wrinkle is a different object.

The length of cloth a fold bends is its arc: the radius times the angle the fold turns through, which is π times the radius for a fold taken through a half-turn. A poplin folded at its warp’s handover radius, 4.22 millimetres, bends 13.3 millimetres of cloth, and at twenty-two picks to the centimetre that is 29 picks. Folded across the weft at the weft’s handover radius, 1.85 millimetres, it bends 5.8 millimetres, and at thirty-two ends to the centimetre that is 19 ends. The fold that matters to a wrinkle spans two or three dozen threads, not two or three.

The least crimp a fold can find in four weaves, against the length of its arc. For four weaves, the least crimp any placement of a fold can find under its arc, as a share of the weave's average, against the arc's length in thread pitches from 1 to 40, with each turn's crimp spread over the pitch it sits in: 2/1 twill 1.00 at 3, 0.90 at 10, 0.98 at 20, 1.00 at 30; 3/1 twill 0.67 at 3, 0.80 at 10, 1.00 at 20, 0.93 at 30; 5-end satin 0.00 at 3, 1.00 at 10, 1.00 at 20, 1.00 at 30; 8-end satin 0.00 at 3, 0.80 at 10, 0.80 at 20, 0.80 at 30. A plain weave would be a line at 1 everywhere. Each curve reaches 1 exactly at whole repeats and dips between them, by less the more repeats the arc holds. 2/1 twill first finds crimp everywhere at 1.1 pitches; 3/1 twill first finds crimp everywhere at 2.1 pitches; 5-end satin first finds crimp everywhere at 3.1 pitches; 8-end satin first finds crimp everywhere at 6.1 pitches. The curves are the same for a fold across the warp and across the weft, since every twill and satin presents the same runs both ways. poplin, across the warp is marked at 29.2 pitches; across the weft is marked at 18.6 pitches. What the picture cannot show is how far along a thread friction would let a fold borrow crimp from outside its arc.
Fig. 4 The least crimp any placement of a fold can find under its arc, as a share of the weave’s average, against the arc’s length in thread pitches, with each turn’s crimp spread over the pitch it sits in. Below a few pitches a satin’s fold can find nothing; by twenty pitches every twill finds at least nine tenths of the average and an eight-end satin four fifths. Each curve returns to one at every whole repeat. The poplin’s two handover arcs are ruled.

The curves say how much of the average the worst-placed fold can count on. Over a short arc a weave can fail a fold entirely: an eight-end satin’s ends run straight for six clear pitches between the turns at either end of a float, so a fold whose arc is shorter than that can land on nothing at all, and a five-end satin’s for three. Over a long arc the shortfall shrinks, because the arc holds several repeats and only the part-repeat at its end can be unlucky. A 2/1 twill’s worst fold finds 90 per cent of the average at ten pitches and 98 at twenty; a 3/1 twill’s 80 and 100; a five-end satin’s returns to 100 at every multiple of five.

The eight-end satin is the slow one. Its two turns sit side by side at the ends of a seven-pick float and then nothing comes for six pitches, so its crimp arrives in clumps an eight-pitch repeat apart, and a window that holds three and a half repeats can hold three clumps or four. Its worst fold finds 80 per cent of the average at ten pitches, at twenty and again at thirty.

The curves are drawn once, and the drawing serves for both directions, because every weave on it is one of the symmetric ones.

What the weave does to the poplin’s two radii

To turn a share into a radius takes a cloth, and the cloth here is the poplin, with a caveat that has to be stated before any number is read: the locus that gives the poplin its two budgets is a plain weave’s. The same yarns at the same setts woven as a twill would sit at a different place on a different locus, and that locus is not computed here. What can be computed is the multiplier each weave’s turns put on the plain cloth’s radii — the part of the question the weave alone decides.

Both handover radii of the poplin, woven in eight weaves. For the poplin — its thickness and its two crimp budgets held at the plain construction's — the radius at which the worst-placed fold first stretches a thread, across the warp (circles) and across the weft (squares), in eight weaves, for a fold through 180°: plain 4.22 and 1.85 mm, a window of 2.28×; 2/1 twill 4.26 and 1.93 mm, a window of 2.21×; 2/2 twill 4.28 and 1.92 mm, a window of 2.23×; 3/1 twill 4.43 and 1.92 mm, a window of 2.31×; 5-end satin 4.29 and 1.93 mm, a window of 2.22×; 8-end satin 4.53 and 2.25 mm, a window of 2.01×; 2/2 warp rib 4.28 and 1.85 mm, a window of 2.32×; 2/2 weft rib 4.22 and 1.92 mm, a window of 2.20×. The dashed lines are the plain weave's radii, 4.22 and 1.85 mm. The widest window is the 2/2 warp rib's and the narrowest the 8-end satin's. The budgets of these yarns woven as a twill or satin are not computed — only the multiplier each weave's turns put on the plain cloth's radii — so the picture shows what the weave adds, not a twill cloth's own radii.
Fig. 5 The poplin’s two handover radii in eight weaves, at the worst placement, with the plain construction’s thickness and budgets: across the warp (circles) and across the weft (squares). The plain weave’s radii, 4.22 and 1.85 millimetres, are dashed. Every weave moves both radii outwards; the multiple beside each row, the larger radius over the smaller, is the window, and it runs from 2.01 for an eight-end satin to 2.32 for a warp rib against the plain weave’s 2.28.

The worst-placed fold hands over sooner than the average one, because the arc it lands on holds less crimp than the budget promises, so every weave moves both radii outwards. The size of the move is the size of the shortfall at that arc. Across the warp the poplin’s arc is 29 picks long, which is long enough that the twills and the five-end satin lose one to two per cent: the radius moves from 4.22 millimetres to 4.26 for a 2/1 twill and 4.29 for a five-end satin, and to 4.43 for the 3/1 twill, whose arc happens to end where its shortfall peaks. Across the weft the arc is 19 ends, and the move is four per cent: 1.85 to 1.93.

The window therefore narrows. The direction that hands over second, across the weft, has the shorter arc and meets more of the weave’s unevenness, so its radius rises by more, and the gap between the two closes: 2.21 for the 2/1 twill, 2.22 for the five-end satin, against the plain weave’s 2.28. The eight-end satin narrows it furthest, to 2.01, because its slow clumping costs the short arc a fifth of its crimp and the long one a fifteenth.

Only two weaves widen it, and both by a hair. The 3/1 twill’s long arc lands on one of its own shortfall peaks and takes the warp’s radius to 4.43, so its window reads 2.31. The warp rib widens it to 2.32 for the reason a rib has: its ends are uneven and its picks are not, so it lifts the warp’s radius and leaves the weft’s exactly where the plain weave had it. The weft rib, the same weave turned through a right angle, lifts the weft’s instead and narrows the window to 2.20.

The weft-dense cloth gives the same answer

The prediction distinguished a warp-faced gabardine from a weft-faced sateen, and the face has turned out not to matter. What a weft-faced sateen usually is, though, is weft-dense — set closer in its picks than its ends — and that changes which direction’s budget is the larger and which arc the longer. The cleanest weft-dense cloth to put beside the poplin is the poplin itself with its two systems exchanged: the same two yarns and the same two setts, the fine dense one now running across.

Both handover radii of the poplin with its two systems exchanged, woven in eight weaves. For the poplin with its two systems exchanged — its thickness and its two crimp budgets held at the plain construction's — the radius at which the worst-placed fold first stretches a thread, across the warp (circles) and across the weft (squares), in eight weaves, for a fold through 180°: plain 1.63 and 3.72 mm, a window of 2.28×; 2/1 twill 1.68 and 3.78 mm, a window of 2.25×; 2/2 twill 1.71 and 3.74 mm, a window of 2.19×; 3/1 twill 1.81 and 3.88 mm, a window of 2.15×; 5-end satin 1.84 and 4.09 mm, a window of 2.22×; 8-end satin 2.20 and 4.40 mm, a window of 2.00×; 2/2 warp rib 1.71 and 3.72 mm, a window of 2.17×; 2/2 weft rib 1.63 and 3.74 mm, a window of 2.30×. The dashed lines are the plain weave's radii, 1.63 and 3.72 mm. The widest window is the 2/2 weft rib's and the narrowest the 8-end satin's. The budgets of these yarns woven as a twill or satin are not computed — only the multiplier each weave's turns put on the plain cloth's radii — so the picture shows what the weave adds, not a twill cloth's own radii.
Fig. 6 The same eight weaves on the poplin with its two systems exchanged, so that the dense system runs across the cloth. The plain weave’s radii swap ends — 1.63 millimetres across the warp and 3.72 across the weft, a window of 2.28 the other way round. The twills and satins again move both radii out and the window narrows, to 2.00 for the eight-end satin; the warp rib now narrows it, to 2.17, and the weft rib widens it.

Exchanged, the plain cloth hands over across the weft at 3.72 millimetres and across the warp at 1.63, and the window is the same 2.28 pointing the other way. Woven as an eight-end satin it narrows to 2.00, as it did on the poplin; the five-end satin to 2.22 and the 2/1 twill to 2.25. In both cloths the satin cancels part of the budgets’ asymmetry, so the sateen half of the prediction comes out right in sign — but for a reason that applies equally to the gabardine. The weave narrows whichever window the budgets open, because the direction with the smaller radius has the shorter arc, and the shorter arc sees more of the weave.

The two ribs swap roles, as they must. The warp rib’s uneven direction is now the one with the shorter arc, so it narrows the exchanged cloth’s window to 2.17; the weft rib, uneven across, widens it to 2.30. A rib widens a window only when its uneven direction is the one that hands over first.

A shallow wrinkle sees more of the weave

Every number so far takes the fold through a half-turn. A wrinkle in a sleeve is not always folded flat; it may turn through a right angle or less, and a fold that turns through less has a shorter arc at the same radius. A shorter arc holds fewer repeats and meets more of the weave.

The poplin's window in four weaves, against how far the fold turns. The ratio of the poplin's two worst-placed handover radii, across the warp over across the weft, for four weaves and folds turning from 15° to 180°, with the plain construction's budgets: 2/1 twill 2.24 at 45°, 2.10 at 90°, 2.21 at 180°; 3/1 twill 1.76 at 45°, 2.09 at 90°, 2.31 at 180°; 5-end satin 2.67 at 45°, 2.22 at 90°, 2.22 at 180°; 8-end satin 1.59 at 45°, 1.59 at 90°, 2.01 at 180°. The plain weave's window is 2.28 at every angle, dashed. Over the whole range the 2/1 twill runs from 1.88 to 2.55, either side of the plain weave's; the 3/1 twill runs from 1.69 to 2.66, either side of the plain weave's; the 5-end satin runs from 1.60 to 2.71, either side of the plain weave's; the 8-end satin runs from 1.50 to 2.80, either side of the plain weave's. A fold that turns less has a shorter arc at the same radius and so spans fewer threads, which is what moves the window; which way it moves is decided by where each direction's arc falls against the repeat. What the picture cannot show is the angle a real wrinkle turns through, which is not one number.
Fig. 7 The poplin’s window — warp radius over weft radius at the worst placement — for four weaves, against how far the fold turns, from 15 to 180 degrees. The plain weave’s 2.28 is dashed and does not move. Each weave’s points fall either side of it, jumping as each direction’s arc passes a whole repeat: at a right angle the eight-end satin’s window is 1.59, and at 45 degrees the five-end satin’s is 2.67.

At a right angle the arcs are half as long — fifteen picks and ten ends at the poplin’s two radii — and the eight-end satin’s window falls to 1.59: its fold across the weft has lost so much to the satin’s clumps that its radius has moved out most of the way towards the warp’s. The other weaves’ points scatter either side of the plain weave’s line. A five-end satin’s window is 2.67 at 45 degrees, wider than the plain weave’s, because at that angle the warp’s arc ends just short of a whole repeat and the weft’s just past one; five degrees later the two have traded places.

That scatter is the real content of the answer. Which way a weave moves a cloth’s window is decided by where each direction’s arc ends against the repeat, and that depends on the radius, the angle and the sett together. It is not a property of the weave’s face, and over the range of angles a real wrinkle takes it has no fixed sign: between 15 and 180 degrees the 2/1 twill’s window runs from 1.88 to 2.55 and the 3/1 twill’s from 1.69 to 2.66, either side of the plain weave’s 2.28. The eight-end satin swings furthest, from 1.50 to 2.80, because its unevenness is large at every arc shorter than a few repeats; it sits below the plain weave at 45, 90 and 180 degrees and well above it between about 50 and 75.

A crease is where the weave decides everything

The gentle folds above are where the weave adds a few per cent. A pressed crease is folded to a radius of the order of the cloth’s own thickness, a third of a millimetre on the poplin, and its arc is one millimetre — two or three threads. On that arc the weave is everything: a five-end satin’s worst fold finds nothing and a 3/1 twill’s two thirds, and at a fold one thread wide a warp rib leaves half its ends bare across the warp and none of its picks across the weft. That is the scale the account of how a crease forms was written at, and the one at which a rib really does crease one way more readily than the other.

It matters less than it sounds, because a crease that sharp has already left the crimp behind. At a radius of a third of a millimetre the poplin’s outer face needs half its length again, and the crimp, even all of it, supplies four to nine per cent. Whether a fold finds that small share or none, the fibres take nearly all of the rest — which is why the fibres decide a crease and the cloth only a wrinkle. The weave’s direction acts where it is drowned out, and fades where the budgets’ direction acts.

How the shares and the radii were computed

The turns are read from the binary matrix of each weave: a warp end turns between two picks where its face changes, a pick between two ends, reading the repeat cyclically. Each turn’s crimp is spread over the one pitch it sits in, as a box of crimp one thread wide centred on the change of face. The first version of the count put each turn’s crimp at a point, and it made the plain weave uneven — a window two and a half pitches long held two point turns or three depending on where it started — so every weave carried a few per cent of unevenness that belonged to the counting. With boxes the plain weave’s crimp tiles its threads exactly and every remaining shortfall is the weave’s.

The worst share is the least crimp under a window of the fold’s arc, over every thread and every continuous position, as a share of the average. The crimp under a window is piecewise linear in its position, so its least value is reached where an end of the window meets the edge of a box, and every such edge is evaluated rather than sampled. A direction’s worst radius is the largest at which the fold’s demand, half the thickness over the radius, exceeds that direction’s budget times the worst share at that radius’s arc; the arc is the radius times the fold’s angle, in the bent system’s pitch. Scanned from above, because the share is not monotone.

The thickness and budgets are the plain poplin’s, from the locus that gave one budget for two directions; the exchanged poplin is the same construction with its two systems swapped. The census enumerates every three-by-three and four-by-four matrix, keeps those that hang together, and compares the multiset of each column’s turn pattern with each row’s, up to rotation and reading backwards.

Required of the count, and checked: that every regular twill and satin and every one’s reverse face present the same runs both ways and a warp rib does not; that a plain weave holds its average under every arc; that every weave holds exactly its average under a whole number of repeats; that an eight-end satin’s fold finds crimp everywhere only once its arc spans the six clear pitches of its float; and that with an even share both radii are the plain construction’s own. A fold through more than a half-turn, an arc of no length, a direction with no budget and a census too large to enumerate are refused.

What the count leaves out

A twill’s own budgets. The poplin’s budgets are a plain weave’s, and a twill of the same yarns at the same setts would sit elsewhere: fewer turns mean less crimp to start with, and a looser jam. Every radius above is the plain cloth’s times the weave’s multiplier, and a twill’s real radii could be larger or smaller by more than the multiplier moves them. The direction a gabardine keeps its wrinkles in is decided there, in its two budgets, and this count does not reach them.

Borrowing. A fold is allowed only the crimp under its own arc. Friction at the crossings is what stops a fold drawing crimp along a thread from beyond the bend, and a thread can slide some way through its neighbours before friction stops it; every shortfall here is therefore an upper bound, and the long arcs at which wrinkles live are exactly where a little borrowing would erase what is left.

The kernel. A box one pitch wide is the simplest honest spreading of a turn’s crimp. A real crimp wave has a shape, and on a float’s end it may extend further along the float than into the interlacing; a different kernel moves every share at short arcs and few at long ones.

One fold at a time. A real wrinkle is a field of folds of many radii and angles, and the angle figure says the weave’s effect on the window jumps with both. What a crumpled specimen shows is an average over that field, and a wrinkle cannot settle what a crease settles for related reasons; nothing here averages it.

Who found what

Peirce’s locus, from 1937, is where the two budgets come from, and that crimp is made at the interlacings is as old as the geometry of cloth. That a crease along a twill line differs from one across it is known from the way twills wear, and a crease across a twill fades in dashes for a reason of its own. What is added here is that a fold counts turns without regard to face, so a twill, its reverse and the same twill folded the other way are one weave to a fold; that every regular twill and satin is symmetric in this sense while 72 per cent of four-by-four weaves are not; that at a wrinkle’s arc the weave lifts both handover radii by a few per cent and an eight-end satin’s by up to a fifth; and that it narrows the window between them in both a warp-dense and a weft-dense poplin, against the expectation that a warp-faced cloth would widen it.

Still open: a twill’s own two budgets

The gabardine question has moved rather than closed. If its floats cannot point its wrinkles, its budgets must, and its budgets are the ends of a locus that has only been computed for a plain weave. A twill’s ends interlace less often and take up less crimp, and they jam against fewer neighbours, so both ends of its locus move: a balanced twill need not have two equal faces found the draft blind to differences the yarn makes, and the locus may be the place a twill’s construction finally shows.

The calculation is Peirce’s geometry with a float in it — a thread that runs straight over several crossings, then bends round one — taken to the jam in each direction. Its prediction would be a pair of budgets for a 2/1 twill at a gabardine’s setts, and the window between them. If the window is wider than a plain cloth’s at the same setts, the gabardine’s horizontal wrinkles belong to its construction after all, by way of the locus; if it is the same, they belong to the setts and the weave was never part of the answer.

Shares its objects with

Essays naming at least two of the same things, that neither author linked.

Named objects

A flat tag is an object no other essay names yet.

BalanceCreaseFloatFold radiusInterchange budgetSatinTwillWrinkle