Textile Structure
how cloth is put together
After the loom

Singeing is the cheapest change to a surface

Pass a cloth through a flame fast enough and it loses under one per cent of its mass. What it loses is the part of itself that was doing most of the touching, and lustre, friction, printability, pilling and measured cover all move at once.
14 min read 6 figures Structure before fibreA mechanism, not a material

Worth reading first: The hairs are what touch · Only a float can be raised · Calendering is the cloth arriving at the other model.

Almost every finishing operation this collection has looked at spends something substantial. Calendering spends the cloth’s compressibility, permanently. Raising spends its strength. Milling spends its dimensions. Each is a real transaction with a real cost.

Singeing costs almost nothing. The cloth passes over a gas flame at fifty to a hundred metres a minute; the protruding fibre ends burn and the body of the yarn, which has thermal mass and no oxygen around it, does not. Under one per cent of the fabric’s weight is gone, no thread has been moved, no crimp has changed, and the weave matrix is exactly what it was.

And the cloth is different in five measurable ways at once.

The two diameters of a 20 tex yarn. The pressure inside a twisted yarn is zero at its surface, so the outermost fibres are held by nothing but their own buried ends and some of them stand off as loops and ends. A yarn therefore has two diameters: a mass diameter of 167.1 µm, which is a volume divided by a length and is the one every other calculation on this site uses, and a contact diameter of 217.1 µm, which is what a neighbouring thread, a finger or an air stream meets. The gap is a hair layer of 25.0 µm on each side and it is measured, not computed — nothing here predicts hairiness. What is computed is the consequence, and it is 29.9% of the diameter every cover factor on this site was built from.
Fig. 1 The yarn as it arrives at the singeing frame. Its mass diameter is 167 µm and its contact diameter 217, the difference being the fibres held by nothing because the pressure in a twisted yarn falls to zero at its surface. Everything the flame removes is in the outer thirty per cent, and almost none of the mass is.

The claim

Singeing removes the part of a yarn that is doing the touching, which is a large fraction of its surface and a very small fraction of its mass.

The asymmetry is arithmetic rather than rhetoric. Take an ordinary ring-spun 20 tex cotton:

  • Its total protruding fibre length is about five centimetres per centimetre of yarn, which is what a hairiness instrument reports. At 0.17 tex a fibre that is 0.85 µg per centimetre against the yarn’s 20, so roughly four per cent of the yarn’s mass is standing outside its own body.
  • Singeing does not remove all of it. The flame reaches what stands clear — the long hairs, above a few tenths of a millimetre — and that is a small part of the length and a smaller part of the number. Reported singeing losses are between a fifth and one per cent of fabric weight, which is consistent with removing perhaps a fifth of the protruding material.
  • But the long hairs are the ones that decide almost everything, because they are the ones that reach.

That last point is the whole mechanism. What matters about a hair layer is its outer radius, and what matters about its mass is its inner bulk, and the two are different fibres.

The five things that move

Lustre. A surface covered in fibre ends pointing in every direction scatters light diffusely; a surface of parallel cylinders reflects it specularly. This collection has computed the geometric part of that for floats and declined the rest, and the same refusal applies here — nothing here is a reflectance model. What can be said is that singeing converts the surface from the first kind to the second, which is why a singed cotton takes a lustre from calendering and an unsinged one does not.

Friction. Friction is a hair property: two yarns crossing meet through their hair layers first, and a yarn-on-yarn coefficient measures two brushes shearing. Removing the brush changes the number without changing the fibre, which is one of the reasons reported friction ranges for a single fibre are as wide as they are.

Measured cover. The gap between the cover a cloth has and the cover a measurement meets is 2h times the sett, and singeing reduces h.

The cover a cloth has and the cover it shows. Cover is sett times diameter, and a yarn has two diameters. The lower line is the cover the construction gives, from the mass diameter of 167.1 µm; the upper is the cover a measurement meets, from the contact diameter of 217.1 µm. The gap between them is 2h times the sett — an absolute addition rather than a proportional one, so it grows in exact proportion to the sett and is asserted to do so. At 8 ends per centimetre it is 0.040 of cover and at 40 it is 0.200. A close cloth is further from its own arithmetic than an open one, which is the wrong way round from every other correction in this collection: the usual shape is a correction that matters when a quantity is small.
Fig. 2 The gap before singeing: 0.12 of cover at 24 ends per centimetre and 0.20 at 40, growing in exact proportion to the sett. Singeing closes most of it, so a singed cloth measures more open than the same cloth did an hour earlier while being structurally identical. Anyone comparing a construction against a measured cover has to know which side of the singeing frame the sample came from.

Printability. A printed line on a hairy cloth is blurred by the hairs, which wick dye out beyond the line and stand above the surface where the screen or the roller cannot reach them evenly. The resolution limit is set by the hair length, so removing hairs of a quarter of a millimetre buys a quarter of a millimetre of definition — which in a fine print is the difference between a line and a smudge. Singeing before printing is universal for exactly this reason.

Pilling. A pill is protruding fibres tangled into a ball and held to the cloth by a few fibres still anchored. Removing the protruding fibres removes the raw material. Singeing does not stop pilling — new fibre ends work their way out of the yarn in wear — but it removes the head start.

The two diameters of a 20 tex yarn. The pressure inside a twisted yarn is zero at its surface, so the outermost fibres are held by nothing but their own buried ends and some of them stand off as loops and ends. A yarn therefore has two diameters: a mass diameter of 167.1 µm, which is a volume divided by a length and is the one every other calculation on this site uses, and a contact diameter of 175.1 µm, which is what a neighbouring thread, a finger or an air stream meets. The gap is a hair layer of 4.0 µm on each side and it is measured, not computed — nothing here predicts hairiness. What is computed is the consequence, and it is 4.8% of the diameter every cover factor on this site was built from.
Fig. 3 The same yarn after singeing, with the hair layer reduced to a few microns. The contact diameter is now 175 µm against a mass diameter of 167 — an excess of five per cent instead of thirty. Structurally nothing has happened; every number this collection computes about the cloth is unchanged; and the yarn has become, for the first time, approximately the cylinder the arithmetic has always assumed it was.

The operation that does the opposite

It is worth putting singeing beside its inverse, because the pair makes the mechanism obvious.

The two diameters of a 20 tex yarn. The pressure inside a twisted yarn is zero at its surface, so the outermost fibres are held by nothing but their own buried ends and some of them stand off as loops and ends. A yarn therefore has two diameters: a mass diameter of 167.1 µm, which is a volume divided by a length and is the one every other calculation on this site uses, and a contact diameter of 191.1 µm, which is what a neighbouring thread, a finger or an air stream meets. The gap is a hair layer of 12.0 µm on each side and it is measured, not computed — nothing here predicts hairiness. What is computed is the consequence, and it is 14.4% of the diameter every cover factor on this site was built from.
Fig. 4 A part-singed surface, between the two above. The hair length is the only quantity that has moved and the yarn underneath is untouched — which is the sense in which singeing is the cheapest change available: it removes a layer that was never load-bearing.

The asymmetry between them is exactly the asymmetry of the pressure field. Raising has to pull fibres out against the grip that is holding them, so it damages the yarn; singeing removes fibres that are already outside the region where any grip exists. One operation fights the pressure field and the other exploits its zero.

That is also why raising needs floats and singeing does not care about the weave at all. A raising machine needs somewhere to get hold; a flame needs nothing but a surface.

Why the flame and not a blade

There is an obvious alternative and its failure is instructive: shear the surface instead of burning it. Cropping machines exist, they are the standard finish for a woollen cloth, and they are not used on cotton for the hair layer.

The reason is geometric. A blade cuts what stands proud of a plane, so it removes the outer part of every hair and leaves a stubble of uniform height. A flame removes what is thin enough to reach ignition before the heat runs away into the yarn, which is a different criterion: it takes the fine, isolated, well-exposed hairs preferentially and leaves the thick clumps. Since the fine isolated hairs are exactly the ones that reach furthest and scatter most, the flame’s criterion is nearer the one that matters.

A blade also has a floor set by the cloth’s own unevenness. Shearing close enough to remove a quarter-millimetre hair on a cloth whose surface varies by more than that means cutting into the threads, which is why cropping is used where a nap has to be levelled — a nap being a surface deliberately built up to a height a blade can work at — and not where a hair layer has to be removed.

The pressure a 23° 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 23° the pressure on the axis is 7.6 per cent of the core fibre's own axial stress and the mean over the section is 3.61 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.
Fig. 5 The reason there is anything to remove. The pressure holding the fibres is greatest on the axis and exactly zero at the surface, so hairs are not a defect of spinning but a boundary condition of twisting. No process that acts on the inside of a yarn can prevent them, which is why every remedy is either a surface operation like this one, a covering operation like sizing, or a construction that gives the zero less to work with.

What a singed cloth is, for the arithmetic

The practical upshot for anybody computing from a construction is small and sharp.

Singeing moves the cloth’s contact diameter towards its mass diameter and changes nothing else. So a singed cloth is the one this collection’s arithmetic actually describes: cover, jam, hole size and thread spacing all computed from the mass diameter are nearer the truth after singeing than before.

That has a slightly comic consequence. The collection has been computing, all along, the properties of a cloth that has been through an operation it has never mentioned — and the agreement between its arithmetic and measured fabrics is partly because most measured fabrics have been singed, since almost every cotton cloth intended to be printed, dyed to a clear shade or given a lustre goes through the frame.

The exceptions are the ones where the arithmetic should be expected to do worst. A flannel, a raised cloth, a towelling, a knitted jersey and anything with a soft handle are all deliberately hairy, and for those the contact diameter is what governs and the mass diameter is a fiction.

How much of a fibre a 23° twist can hold. One staple fibre, drawn to scale along its own length. A fibre end embedded in the yarn is held by friction under the twist's radial pressure and breaks rather than slips once enough of it is buried. Writing the two out and putting the pressure at the value it has when the yarn is at the point of breaking, the fibre's strength cancels and so does the load on the yarn: what is left is a critical length of 5.52 mm — 462 fibre diameters — against a staple of 28 mm. So 19.7% of the fibre is spent being gripped and the rest of it can be broken, which gives a cohesion factor of 0.902. The scale carries the one measured number in this argument, a contact efficiency of 0.05, and the caption of every figure that depends on it says so.
Fig. 6 And what the singeing does not touch. The twist holding the fibres into the yarn is unchanged, so nothing about the yarn’s strength or its diameter moves — the change is entirely in the population standing off the surface, which is why it costs so little and buys so much.

The scaling, which says who needs it

Two of the essay’s quantities scale in opposite directions with the yarn, which decides where singeing is worth its cost.

The relative excess goes as the reciprocal of the diameter. The same absolute hair layer is fifty-five per cent of a 6 tex yarn’s diameter and seventeen per cent of a 60 tex weft’s, so the finest cloths are the ones most changed by singeing.

The number of hairs goes as the fibre ends per unit length, which is the fibre count divided by the staple. A coarse yarn has more fibres and therefore more ends, and a short-stapled cotton has more ends than a long-stapled one at the same count.

The two together say that singeing matters most for fine cloths of short-stapled cotton, which is exactly the category — poplins, voiles, lawns, printed cottons — where it is universal, and least for coarse long-stapled goods, where it is often skipped.

Reach per unit mass goes as the square of the fineness

The essay’s asymmetry — a large fraction of the surface and a very small fraction of the mass — has an exact form, and writing it out says which yarns singeing is worth most on for a reason the scaling section does not reach.

A protruding hair of length h and fibre diameter df has a mass proportional to h times df squared, and a reach of exactly h. So reach per unit of mass goes as one over the fibre’s diameter squared, and a fibre half as thick reaches four times as far for the same weight of material standing outside the yarn.

That is a stronger dependence than anything else in this essay, and it puts the fine-fibred yarns in a category of their own. A microfibre or a fine-wool yarn carries a hair layer that is large in every dimension that decides the surface and negligible in the one that appears on a weighbridge — so the singeing loss is smaller than for a coarse yarn and the change to the fabric is larger. The two move in opposite directions with the same variable, which is unusual and is the reason the operation looks so much like a free lunch on exactly the goods where it is most needed.

It also says something about what a hairiness measurement is reading. An instrument that reports total protruding length is reading a quantity proportional to reach and blind to fibre fineness; one that reported protruding mass would rank the same yarns quite differently. Two yarns of equal hairiness by length can differ fourfold in the mass standing outside them, and only the first of those numbers is the one that decides how the cloth behaves.

Why the consistency check does not close

The essay’s own arithmetic puts four per cent of a yarn’s mass outside its body and the reported singeing losses at under one per cent, and the gap is left standing as an unclosed check. It is worth saying what would close it, because the answer is a distribution rather than a correction.

The flame does not remove hairs; it removes hairs above a threshold. What stands clear enough to reach ignition before the heat runs into the yarn is a few tenths of a millimetre and up, and everything shorter is protected by its own neighbours. So the fraction removed is the fraction of the protruding material lying beyond a cut in a length distribution, and that fraction depends very steeply on where the cut sits relative to the distribution’s own decay length.

A hair population that falls off over a few tenths of a millimetre and a threshold at a few tenths of a millimetre put the cut right in the middle of the steep part, which is exactly where a fifth is a perfectly ordinary answer and where a small change in the flame’s reach moves it a long way. The check does not close because it cannot close without the population, and it would be a worse check if it did — a ratio that came out at one from two independently reported numbers would be evidence that one of them had been fitted to the other.

Where it sits in the finishing sequence

The position of singeing in a finishing route is decided almost entirely by the arithmetic above, and it is worth setting down because each placement is a consequence rather than a convention.

Before desizing and bleaching, because the size that was applied to lay the warp’s hairs down for weaving is still on the cloth, and a sized yarn’s hairs are glued to its body where a flame cannot get at them. Singeing a sized cloth removes almost nothing. So the order is: weave sized, desize, singe.

Before dyeing, because a protruding fibre takes dye and stands out of the surface, so a hairy cloth dyed to a deep shade looks frosted where the hairs catch the light differently from the body. This is the same argument as the printing one with a lower resolution requirement.

Before calendering, because a calender spends the cloth’s compression for good and what it presses is whatever is on the surface. Pressing a hair layer flat gives a temporary lustre that the first wash removes; pressing a singed surface gives one that survives, because it is the threads themselves that have been flattened.

And never after raising, which is the operation whose whole purpose is the layer this one removes.

The sequence is a small worked example of a general point about finishing routes: they look like tradition and they are usually the unique order in which each operation has something to work on.

The cheapest change, priced properly

The essay’s title is a claim about a ratio and it is worth stating the ratio.

Singeing removes between a fifth and one per cent of a cloth’s mass and no part of its structure. Against that it moves the contact diameter by about a quarter of itself, which changes the measured cover by 0.12 at an ordinary sett, changes the yarn-on-yarn friction by an unquantified but demonstrably large amount, and changes the print resolution by the length of the hairs removed.

Compare that with the other finishing operations this collection has priced. Mercerising changes the packing factor and therefore every diameter in the cloth, and it costs a chemical process and a tensioned frame. Milling shrinks the cloth by a fifth and cannot be undone. Raising removes fibres from the load path.

Singeing is the only one that takes nothing the cloth was using. The material it removes was, by the pressure argument, being held by nothing and carrying nothing — and that is not a happy accident of the process but the definition of what a protruding fibre is.

What was counted, and how

The hair mass is computed from the reported hairiness rather than assumed. Total protruding length times the fibre’s own linear density, against the yarn’s, gives about four per cent — and the comparison with the reported singeing losses of under one per cent is offered as a consistency check that does not quite close, because it implies the flame removes about a fifth of the protruding material and nothing here establishes that independently.

The cover gap is asserted proportional to the sett at every step of the sweep and to twelve decimal places at its ends, which is the form of the claim: an absolute addition to a diameter, multiplied by a sett.

The excess is computed at three counts so that the reciprocal scaling is exhibited rather than stated.

Where the model stops

Nothing here is a combustion model. How much of a hair burns, how far the heat penetrates, what the flame temperature and the cloth speed have to be — all of that is the actual engineering of singeing and none of it is in reach. The essay computes what is removed given that something is removed, and takes the reported weight loss as an input.

The hair layer is again an annulus and is not one. Everything in the rung below about that applies here unchanged, and it is the largest approximation in both.

The five consequences are not equally well supported. The cover claim is arithmetic. The friction claim is a mechanism with a wide measured range. The lustre and printability claims are qualitative, and this collection has declined the thread transmittance precisely so as not to make quantitative optical claims it cannot support.

And there is a sixth consequence that is not about the surface at all. Singeing exposes the cloth to a flame, and the fibres it does not remove are heated. For cotton this is harmless at the speeds used; for a blend containing a thermoplastic it is not, because a melted fibre end forms a hard bead. That is a chemistry question and this collection does not have one.

Where the ladder goes next

Back into the cloth, where the two diameters have to be reconciled with a construction. Every threshold in this collection that is about touching rather than about mass — the jam, the hole, the thickness under a light foot — is computed from the wrong one of the two, and the size of the error is now known.

And sideways into the yarns that avoid the problem by construction rather than by burning. A filament yarn has no fibre ends and no hair layer; a folded yarn traps each single’s surface against its neighbour; and a compact-spun yarn is a ring yarn with the spinning triangle condensed so that fewer fibre ends escape the twist in the first place. All three are ways of arranging for the pressure field’s zero to have less to work with.

Where this goes next

The two diameters of a 20 tex yarn. The pressure inside a twisted yarn is zero at its surface, so the outermost fibres are held by nothing but their own buried ends and some of them stand off as loops and ends. A yarn therefore has two diameters: a mass diameter of 167.1 µm, which is a volume divided by a length and is the one every other calculation on this site uses, and a contact diameter of 217.1 µm, which is what a neighbouring thread, a finger or an air stream meets. The gap is a hair layer of 25.0 µm on each side and it is measured, not computed — nothing here predicts hairiness. What is computed is the consequence, and it is 29.9% of the diameter every cover factor on this site was built from. The hairs are what touch The pressure inside a twisted yarn falls to exactly zero at its surface, so the outermost fibres are held by nothing and some of them stand off. A yarn therefore has two diameters — the one its mass gives and the one a neighbour meets — and only the first is in the arithmetic. What a raising machine can catch in a 2/2 twill. The draft with every weft float long enough for a raising wire to lift marked on it. The teeth need an unsupported length of thread on the surface, so a cloth in which every thread is bound at every crossing offers them nothing at all — which is why a napped fabric is always a twill or a satin and never a plain weave. Only a float can be raised A raising machine drags wire teeth across a cloth and pulls fibre ends up into a nap. The teeth need something to catch, and what they catch is a float — so which fabrics can be napped at all is a question about the matrix, decidable exactly, and the answer over the whole four-by-four census is two. From a calender's line load to a force at one crossing. A sheeting through a nip loaded at 30 N per millimetre of bowl width, with the cloth in contact over 5.0 mm. The pressure is the first divided by the second, 6.00 N/mm², and the force at one crossing is that pressure times the area a crossing owns — the product of the two thread spacings, 0.1374 mm². So the crossing carries 0.824 N, the sections flatten to 2.42 and 2.63, and the cloth thins from 0.388 mm to 0.218 mm. What the drawing cannot show is that two cloths through the same nip are not given the same treatment: the area a crossing owns varies fivefold across this site's table, and it is a factor in the force. A calender spends the compression for good Calendering was described on this site as moving a cloth from one thread-section model to another, which was right and had no number in it because the amount of the move was a free parameter. It is a pressure now — and the same nip setting turns out to give two cloths quite different treatments, because the force at a crossing is the pressure times the area a crossing owns.

What links here

Computed from the collection rather than written here: the essays that point at this one.

Reads more easily once this is understood

Essays that name this one as worth reading first.

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.

Cover factorFibre finenessFrictionHairinessLustreSpecificationStaple lengthYarn diameter