Calendering is the cloth arriving at the other model

Two essays into this site’s foundation, a decision was made that has been carried ever since: when a number depends on what shape a yarn’s cross-section is taken to be, the model is named. Peirce’s geometry treats it as a circle. Kemp’s racetrack treats it as a rectangle with semicircular ends. Every thickness, cover factor and jamming sett on this site says which one produced it.

The two have been presented as competing idealisations of the same object, which is how the literature presents them. They are not.

A calender is two rollers #

The machine is as simple as it sounds. The cloth passes through a nip between two heavy rollers, at least one of them heated, under a pressure of several tonnes across the width. What comes out is thinner, smoother, denser to the eye, and glossier.

What has happened to the yarn is that its section has been flattened. A yarn is not solid — it is fibres with air between them, at a packing factor this site has computed the consequences of — so it deforms readily under pressure applied through its thickness, and the fibres redistribute sideways rather than being compressed.

The cross-sectional area is very nearly conserved, because the fibres are neither destroyed nor squeezed into a smaller volume; they are rearranged. So the yarn becomes wider in the plane of the cloth by whatever factor it becomes thinner through it, and that ratio — width over thickness — is precisely the flatten parameter of Kemp’s racetrack.

What follows immediately #

Three quantities move, and each of them is one this site already computes.

Cover rises. The cover factor is a width divided by a spacing, so a wider yarn at the same sett covers more. In the figure above, a cloth at 2.2 threads per unit width goes from 44 per cent cover to 96 per cent as the flattening runs from 1 to 3 — with exactly the same yarn, in exactly the same fabric, at exactly the same thread count. That is why a calendered cloth is more opaque and why calendering is a route to windproofing — though it does not reach the threshold, and opacity is not cover either.

Thickness falls, by the flattening factor, which is the effect the machine is most obviously for.

And the jamming sett falls, which is the least obvious. A flattened yarn is wider, so the threads run out of room sooner: the closest a cloth could have been woven drops as the flattening rises. That is a statement about a cloth that no longer needs weaving, so it is a curiosity rather than a constraint — but it says something real, which is that a calendered cloth is closer to its own jam than the cloth that went into the machine, and has correspondingly less room to move.

The reframing, and why it is worth an essay #

Here is the claim this essay exists for.

The choice between the two thread models is not a choice about idealisation. It is a choice about which state of the cloth is being described.

A cloth as woven, with round yarns that have never been pressed, is Peirce’s cloth and the circular model is the right one. A cloth as finished, calendered, pressed at a nip under tonnes of load, is Kemp’s cloth and the racetrack is the right one. They do not disagree about a yarn; they describe different yarns, and the machine that turns the first into the second is in every finishing works in the world.

That does not resolve the modelling question so much as relocate it. Kemp’s racetrack was proposed as a better description of real cloth, and the reason it is better is not that it is a more sophisticated idealisation — it is that the cloths available to measure had mostly been through a calender or a press, and the round-yarn cloth was the theoretical one.

A model that fits the data better because the data has been through a machine is not a better model of the underlying object. It is a better model of the object as it is usually met, which is a different and often more useful thing, and the two get conflated whenever the state is left unsaid.

Schreinering, and the other reason for a calender #

There is a variant worth mentioning because it separates two things a plain calender does at once.

A schreiner calender has one roller engraved with very fine diagonal lines — several hundred to the centimetre — set at the angle of the twist in the yarn. It embosses those lines onto the fabric surface, and the result is a high, silky lustre from a cotton cloth.

The mechanism is not flattening. It is that the embossed lines are parallel and at the fibre angle, so light reflecting off the surface reflects coherently in one direction rather than being scattered by the yarn’s own curvature. That is the same mechanism this site computed for satin’s shine — an uninterrupted length of thread reflecting along its own axis — applied at a much finer scale and imposed rather than woven.

So a calender does two separable things: it changes the yarn’s cross-section, which changes cover and thickness and is what this essay computes, and it changes the surface’s optical geometry, which is not computed here at all and belongs with the site’s reflection arguments.

The distinction matters because the two have different permanence. Flattening is fairly durable and lustre is not. A schreinered finish washes out; the cloth’s cover does not go back.

What was counted, and how #

calenderSeries calls racetrack() at each flattening and asserts two things that could fail.

The area is conserved across the whole series, to nine decimal places. racetrack() computes the thickness b from the area and the flattening by solving area = πb²/4 + (a − b)b, so equal areas is what the implementation is for — and asserting it catches an algebra error in that solve, which is the only place in this family a mistake would produce a smoothly wrong picture.

A calendered yarn is wider and thinner, both, which would fail if the flattening had been applied to the wrong axis. That is a mistake with a fifty per cent prior and no visual signature: a series of ellipses getting taller instead of wider looks exactly as plausible as the correct one.

The cover figures come from the width and the stated sett; the jamming setts from jammedSett, which this site has carried since the setting essays and which computes the closest possible sett by each model at the weave angle of 60° that the closure condition forces.

The permanence question, separated #

Everything a calender does relaxes, and the rates differ enough that they are separate properties rather than one.

The flattening is the most durable, because it is a mechanical rearrangement of fibres inside a yarn and there is no strong restoring force returning them. It relaxes slowly with washing and wear.

The surface lustre is the least durable, because it is a micro-geometry imposed on the outermost fibres and any abrasion disturbs it. A schreinered cotton loses most of its lustre in a few launderings.

And for a thermoplastic fibre both can be made permanent, by running the nip above the fibre’s softening point so that the deformation is set into the polymer rather than merely imposed on the assembly. That is why calendered synthetics hold their finish and calendered cottons do not, and it is a fibre property rather than a machine one.

Where the model stops #

Area conservation is an idealisation. A yarn under a calender nip does consolidate somewhat — air is expelled and the packing factor rises — so a real flattened yarn has a slightly smaller area than the one that went in. The model has none of that.

The racetrack is a shape, not a mechanics. Nothing here computes what pressure produces what flattening, or how much recovers when the load comes off. Real flattening is partly elastic and a calendered cloth relaxes some of it, especially when wetted, which is why calendered finishes are refreshed rather than permanent.

The fabric is not modelled at all. A calender flattens the yarn where it is proud of the surface — at the crimp crowns — much more than where it lies in the interior, so a real cloth’s flattening varies along each thread. Treating it as uniform is a considerable simplification and it is the one that would matter most for a thickness prediction.

And the two thread systems are treated identically. In a real nip the system on the face takes more of the load.

What a designer buys with it #

Calendering is cheap, fast and done to almost everything, so it is worth being clear about what it is for.

Opacity and windproofing, from the cover rise. A cloth that must not be seen through, or must not let air through, can be woven more openly and calendered than woven densely — which is cheaper in yarn and faster on the loom.

Smoothness, which is a handle property and also a functional one: a smoother surface soils less and is easier to clean.

Lustre, from the surface geometry rather than from the flattening.

And thickness control, which matters wherever the fabric has to fit somewhere — a coated substrate, a laminate, a filter housing.

What it does not buy is anything permanent, and this is the honest limit of the operation. Every effect above relaxes with washing and wear, at rates depending on the fibre: a thermoplastic fibre calendered above its glass transition holds the deformation, and a cotton one does not.

What the site should do about having two models #

If the two thread sections are two states rather than two opinions, then this site’s own practice needs a word about it, because it has been quoting both.

The practice has been: name the model with every number. That was right and it remains right, and this essay adds a reading to it. When a figure says “Peirce circular” it is describing an unfinished cloth, and when it says “Kemp racetrack” it is describing a finished one — so the model name has been carrying state information all along without anybody saying so.

That is worth making explicit rather than leaving as a private understanding, and it is a small addition to the site’s fourth invariant. Say which model, and know that saying which model is partly saying which state.

It also resolves an awkwardness the foundation essays left open. Presenting two models side by side and refusing to choose is honest and slightly unsatisfying: a reader reasonably asks which is right. The answer available now is better than a choice. Both are right, about different cloths, and the flattening parameter is the coordinate between them — so a cloth can be placed on the axis rather than assigned to a camp.

Where a woven cloth already sits on the axis #

The reframing places Peirce’s circle at one end of a flattening axis and Kemp’s racetrack at the other, with the calender as the operation between them. That is the right axis and the cloth does not start where the essay puts it.

This collection has measured the starting point, from the other direction. Inverting eight measured fabric thicknesses through a compression energy gives the aspect ratios that reconcile them, and they come out between 1.56 and 2.33, warp and weft — before any machine has touched the cloth. That is the flattening the crossing threads impose on one another, and it is the pressure a loom’s own warp tension supplies rather than anything a finisher does.

So a woven cloth’s yarns are already at a flattening of about 1.9 as they leave the loom, and Kemp’s racetrack — proposed as a description of real cloth — is a description of woven cloth rather than of calendered cloth.

Peirce’s circle, on this reading, describes no fabric at all. It describes a yarn on a bobbin. The moment a thread is woven it is pressed against its crossing partner and is somewhere near the middle of the axis, and every essay on this site that computes from a circular section is computing from a state that exists before the loom and not after it.

Which shrinks what the calender does #

The consequence is a substantial recalibration of the numbers above, and it runs the same way for all of them.

A conserved area makes the width go as the square root of the flattening. So a cloth already at 1.9 has yarns 38 per cent wider than their nominal diameter, and a moderate calendering to 2.4 — the figure this collection computes for a pressed twill — widens them by a further

√(2.4 ÷ 1.9) = 12 per cent.

Twelve, not fifty. The essay’s series from a flattening of one to three takes the cover from 0.44 to 0.76, and the part of that a calender is responsible for is the segment from 1.9 to 3, which takes it from 0.61 to 0.76 — a quarter, not a near-doubling.

The cloth had already travelled two thirds of the way along the axis before it reached the machine, and the machine’s contribution is the last third. That is a much less impressive number for the operation and a much more impressive one for weaving, which turns out to be the larger of the two flattening processes and is never described as one.

And it says which quantities were already wrong #

The same factor reaches every quantity this collection computes from a round diameter, and it reaches them in a direction that is worth stating plainly.

Every cover factor computed from a circular section is low, by about the same 38 per cent, because the width that covers a cloth is the flattened width and not the nominal diameter. A sheeting’s computed 0.52 is nearer 0.72 in the cloth.

That is a large correction and it points the way this collection’s own observations have always pointed. A cloth is more opaque than it is closed is the same discrepancy noticed optically and attributed to the hair layer; a good part of it is simply that the threads are wider than the arithmetic says.

And every jamming sett computed from a circular section is high, by the same factor, because wider threads meet sooner. That runs the same way as the extreme-value correction and compounds with it, so the geometric ceilings this collection quotes are optimistic twice over.

Neither correction changes an ordering, since the factor applies to every draft alike. What it changes is the standing of the absolute numbers, and the honest summary is the one this essay was already reaching for: the model name has been carrying a state all along, and the state the site mostly computes in is the one the cloth leaves behind at the loom.

What a nip does that the model does not capture #

The most serious limitation deserves a section of its own, because it is the difference between a yarn model and a fabric model.

A calender nip closes on a fabric, not on a yarn. What it meets first is the crimp crowns — the places where a thread rises over its crossing partner and stands proudest of the surface — and those take the load before anything else does. So the flattening in a real calendered cloth is concentrated at the crossings and much less in the spans between them.

That has two consequences the model misses. The thickness falls more than the yarn’s own flattening would suggest, because the crowns are exactly what set the thickness. And the cover rises less, because the spans between crossings, which are what cover the interstices, are flattened least.

So the numbers in this essay overstate the cover gain and understate the thickness loss, both in the same direction and for the same reason. A fabric-level model would fix it and would need a load distribution across the crimp profile, which is a substantially larger piece of machinery than anything here.

It is worth noting how cheap the operation is relative to what it changes. A calender adds no material, consumes only heat and pressure, runs at the speed of the rest of the finishing line, and is applied to almost everything. Very few interventions in this subject move a fabric’s cover by fifty per cent for so little, and the reason the effect is not larger still is that the flattening is bounded by what the yarn will take before its fibres begin to break rather than move.

A last practical note about where the operation sits. A calender is placed at the very end of a finishing route, after the stenter and usually after any pre-shrinking, because everything it does is a surface and thickness effect that a later wet operation would undo. That makes it the last machine the cloth meets, which is convenient for the finisher and awkward for anyone measuring the fabric afterwards: a calendered cloth’s thickness, cover and handle are all set by the final ten seconds of a process that took hours, and none of them is recoverable from the construction alone. It is the clearest instance in this field of a fabric property that is a machine setting wearing a fabric’s clothes.

Who found it, and when #

Calendering is old — the mangle is medieval and the principle is unchanged — and Kemp’s racetrack is from 1958, proposed as a description of yarns in cloths that had been through such machines.

That Kemp’s section describes a calendered yarn and Peirce’s a round one is implicit in the literature and rarely said outright. Peirce himself noted that his circular assumption was an idealisation and that real yarns in cloth are flattened by the forces at the crossings, which is a third and separate flattening mechanism — the cloth flattening its own yarns, before any machine touches it.

That mechanism means the reframing in this essay is a simplification of its own. A yarn in a woven cloth is somewhat flattened before it reaches any calender, by the pressure of the crossing threads, and the racetrack’s applicability begins there. The calender takes it much further, and the two models sit at the two ends of a continuum that a cloth is somewhere along.

Where the ladder goes next #

Flattening changes the yarn’s shape. The other route to a wider yarn changes its size, and does it chemically rather than mechanically: mercerisation swells the fibre, which is one number in this site’s diameter calculation and a long list of consequences.