A seam stands proud and wears first

Everything in this ladder has computed a surface at the scale of a thread: crowns a hundred and ninety micrometres above a mid-plane, contact at depths of one to thirty micrometres, wear at a fraction of a thread’s radius. A garment has a feature on it three orders of magnitude larger than any of those, and it is sewn in deliberately.

The claim #

A seam is a step in a garment’s bearing curve two orders of magnitude larger than the cloth’s own relief, so it takes the whole of any flat contact over a couple of per cent of the area.

Three consequences, and the third is a piece of the trade’s practice explained.

Everything this ladder computes applies inside the seam and nowhere else. The crown line, the bearing exponent, the wear concentration — all of it is happening on the seam’s own top ply, and the rest of the garment is out of contact.

The concentration is a product of two concentrations. The seam is two per cent of the area, and within the seam the crowns are a few per cent of the plan, so a rubbing load applied to a garment is being carried on about one part in two thousand of it.

And pressing a seam is a local calender. A seam under an iron gets the treatment a calender gives a whole cloth — flattened threads, a widened flat top, a multiplied specular area — which is exactly the shiny mark a pressed seam leaves, and it is geometric rather than a scorch.

The arithmetic, which is one subtraction #

A garment panel is one ply of fabric. A plain seam allowance is the two panels turned back on themselves plus whatever the stitching adds — two or three plies where the body is one, each of them a cloth whose own thickness is a maximum rather than a mean.

The step is therefore (plies − 1) times the cloth’s thickness, which for a three-ply seam in a 382-micrometre sheeting is 763 micrometres.

Set that beside the depths the rest of this ladder works in. A thickness gauge reaches two micrometres. A hand at five kilopascals reaches five. A firm pinch reaches twenty-eight. The seam’s step is between thirty and four hundred times any of them.

So there is no ordinary contact in which the body cloth participates at all, unless the contacting object is soft enough to drape over the seam — which skin is, and a chair is not, and a table edge is emphatically not.

What fraction of a garment is seam #

The step decides that the seam takes all the contact. How much wear that concentrates depends on how much of the garment is seam, which is a matter of cutting rather than of cloth.

A shirt has perhaps four metres of seam in one and a half square metres of fabric. At ten millimetres of allowance visible on the outside — a felled side seam, a set-in sleeve, a yoke — that is 0.04 square metres, or about two and a half per cent of the garment.

A trouser is worse and a coat is better, and the number is a design output rather than a fabric property. What is not a design output is the consequence: whatever the fraction, the seam is carrying all of the flat contact, so the rubbing per unit area on a seam is one over that fraction — a factor of forty for the shirt.

Which garment features are steps and which are not #

Once the threshold is stated — a few micrometres of contact depth against whatever the feature stands proud by — a garment can be sorted, and the sorting is unforgiving.

Every one of these is a step and takes contact: a seam allowance at two or three plies; a hem at two or three; a cuff, a collar and a waistband at three to six; a pocket edge; a placket; a button and its shank; a zip; a printed or embroidered motif with any relief in it at all.

Almost nothing on a garment is not a step. A single-ply panel with nothing on it is, and that is the only case in which the surface arithmetic of this ladder describes what a garment is actually touching with.

That is a slightly bleak conclusion for a ladder of essays about woven surfaces, and it is the honest one. The surface computed here is what a fabric presents; a garment presents its assembly. The two are related the way a material is related to a structure, and the assembly wins wherever it exists.

Where the fabric surface does govern is in the flat: the back of a shirt against a chair, a sheet against a body, a sleeve against a table over its middle rather than its cuff. Those are real and they are a minority of the contacts a garment has.

The two concentrations, multiplied #

The seam concentrates the contact onto two and a half per cent of the garment. Within the seam, the crowns concentrate it again onto a few per cent of the plan.

At five kilopascals nominal on a two-and-two twill the bearing fraction is 4.4 per cent, so the pressure on the crowns is twenty-three times nominal. Multiplying: a load applied over a garment is carried on about one part in nine hundred of its area, at about nine hundred times the nominal pressure.

That is why an elbow, a cuff edge, a collar fold and a seat seam are where a garment fails, and it is why the failure is so localised. Nothing about the fabric is different there. The geometry is putting the whole of a diffuse load onto a line.

What a wear test is measuring on a made-up garment #

Abrasion standards test fabric, in a circular specimen with no seam in it, and they report a fabric property. Garment wear trials test made-up garments and report where they failed.

The two disagree routinely, and the geometry above says they must. A fabric test measures the surface; a garment test measures the assembly, and the assembly’s steps take all the contact wherever the contacting object is flat.

So the correlation between a Martindale rating and a garment’s service life is expected to be poor, and the reason is not that the test is unrepresentative of the rubbing — it is that the test is representative of a part of the garment that is not being rubbed.

A test that predicted garment life would have to test a seam, which is to say it would have to test a construction rather than a cloth, and it would report a number that belongs to the maker rather than to the mill. That is presumably why nobody does it, and it is a real gap rather than an oversight.

The same reasoning explains a familiar observation: a garment made from a heavier cloth does not last proportionally longer, because its seams are also thicker and the concentration is unchanged. What extends a garment’s life at the seams is a flatter seam — felled, topstitched, pressed open rather than to one side — every one of which reduces the step rather than improving the fabric.

Why a pressed seam shines #

The mark a pressed seam leaves is called seam impression or shine, it is a defect in tailoring, and it is usually explained as glazing or scorching of the fibre.

It is a calender. An iron applies heat and a line load to a narrow strip of cloth, which is what a calender does to a whole one, and the arithmetic is the same: the threads flatten, their sections acquire a flat top, and the specular area of the strip multiplies by the ratio of a flat to an arc.

This collection computes that gain at twenty-four times in width and six per cent in length for a moderate pressing. A pressed seam is a stripe of cloth with twenty-four times the lustre of the fabric beside it, in a fabric that is otherwise uniform — which is exactly a visible line.

Three things follow and all three match practice.

It is worst where the surrounding cloth is most matt, because what is seen is a contrast rather than a brightness. A pressed strip of plain weave beside unpressed plain weave is a large ratio; a pressed satin beside an unpressed one is a smaller one, since the satin was already returning a great deal. That matches the trade’s own view of which fabrics show a seam impression worst, which is worsteds and smooth-faced woollens rather than satins.

It is permanent, because the compression is spent for good and the flattening does not recover.

And a pressing cloth or a strip of card under the allowance prevents it, because it spreads the line load and stops the threads at the edge of the allowance being pressed harder than the rest — which is what tailors do and what they say it is for.

The one seam that is not a step #

There is a construction that removes the step entirely, and it is worth naming because it shows the argument is about geometry and not about seams as such.

A butted seam joins two edges without an overlap: the two panels meet edge to edge and are held by a stitch or a tape spanning the join. There is no allowance, no doubled ply, and no step — the surface across the join is one ply throughout.

That construction exists and is used exactly where the step is the problem: in wetsuits, where an overlapped seam would be felt through the garment; in medical and protective wear, where a proud seam is a pressure point on skin; and in knitted underwear, where the same effect is achieved by knitting the garment in one piece and having no seam at all.

The tailoring trade’s usual answer is different and is the same idea. A felled seam presses the allowance flat and stitches it down, which does not remove the step but halves it and spreads it over twice the width; a pressed-open seam divides one three-ply step into two two-ply ones. Every traditional flat-seam construction is an attack on the same quantity, arrived at by feel long before anybody wrote down a bearing curve.

The seam’s shadow, which is wider than the seam #

The argument above has two cases and a garment lives between them. Against a rigid plane the seam takes everything; against something soft enough to drape, the body cloth participates. What decides how much is a length the trade already measures, so the intermediate case can be computed rather than left as a caveat.

A flexible sheet laid over a step does not fall to the surface at the step’s edge; it bends down over a distance set by its own stiffness against its own weight, which is exactly the bending length from the cantilever test — fifteen to twenty-five millimetres for an ordinary shirting. So the region out of contact is not the seam’s ten millimetres but the seam plus a band of about a bending length on each side.

The out-of-contact region is the seam width plus twice the bending length of whatever is doing the rubbing.

Put the shirt’s four metres of seam through it. The allowance itself is 0.04 square metres; a twenty-millimetre bending length adds a shadow of 2 × 0.02 × 4 = 0.16, so 0.20 square metres of a 1.5-square-metre garment is out of contact — thirteen per cent, not two and a half.

Which cuts the concentration factor from forty to about seven and a half, and does it in the only direction that matters: the body cloth is now carrying something rather than nothing.

Both limits fall out of the same expression, which is the reason to have it. A rigid abradant has an infinite bending length and shadows the whole garment, recovering the essay’s own case. A very limp contacting cloth has almost none and touches everywhere but the allowance. And a garment rubbing against itself sits in the middle, which is most of the wear a garment actually gets — a sleeve against a side, a thigh against a thigh — and is the case neither the fabric test nor the rigid-plane argument describes.

What flattening a seam actually buys #

The essay’s own conclusion is that halving a step which is a hundred times too large changes nothing, and that is right about which cloth is in contact. It understates what a felled seam does, because the height is not the only thing that changes.

A plain seam pressed to one side is a proud band of one allowance width. Felling it folds the allowance under and stitches it down, which spreads the same excess ply over twice the width — and the load is carried over twice the area at half the pressure.

Wear goes roughly as the pressure at first order, so felling doubles the seam’s life without changing the fabric, the thread or the step’s reach. The forty-fold concentration becomes twenty-fold. Pressing a seam open does the same thing differently, dividing one three-ply band into two two-ply bands and again roughly halving the pressure.

So the traditional flat-seam constructions are not failed attempts at removing a step. They are successful attempts at spreading one, and the quantity they move is pressure rather than height. That is why they work, why they were arrived at by feel, and why they have never been explained in terms of the step they conspicuously fail to eliminate.

The design instruction that follows is short and is not the one the step argument suggests on its own. Do not try to make a seam thin; make it wide. A seam half as high and the same width buys nothing. A seam the same height and twice as wide buys a factor of two, and it is available in every sewing room already.

Where the model stops #

Nothing here is a real seam. A seam has stitches in it, the stitches have their own thickness, the plies are not flat against one another, and the allowance may be pressed open, pressed to one side, felled or bound — every one of which changes the step and none of which is computed.

The contacting object is a rigid plane. Skin is not, cloth is not, and a soft object drapes over a seam and does contact the body of the garment. The claim is exact for a hard flat abradant, which is what a wear tester uses and what a table, a desk and a chair arm approximately are.

The two per cent is an estimate from one garment. It is arithmetic on a plausible shirt rather than a measurement, and it is a design quantity that varies by a factor of several between garments.

And the seam’s own strength is a different subject. What holds a thread in a seam and a seam slips before it breaks are about the sewn joint under load; this is about the same joint under a rubbing, and the two failures have nothing in common except their location.

The generalisation #

A step in a surface is a contact concentrator, and its size relative to the surface’s own roughness decides whether the surface matters at all.

That is the transferable statement and it has a threshold in it. Where a step is comparable to the roughness, the two combine and the contact is shared; where a step is a hundred times the roughness, the step takes everything and the roughness is irrelevant.

Almost every made object has features of both kinds — a seam, a hem, a fold, a button, a printed edge, a label — and the rule for reading them is a comparison of scales. Compute the surface’s own contact depth first, which for cloth is a few micrometres, and then ask which features of the object are larger than that. Everything larger is carrying the load; everything smaller is not there.

One more number makes the point about scale. A felled seam of two plies rather than three has a step of 382 micrometres instead of 763 — halved, and still fifty times the depth at which the body cloth first touches. Halving a step that is a hundred times too large changes nothing about which part of the garment is in contact. To get the body cloth into play the step would have to come down to a few micrometres, which is a butted seam or nothing.

Who found it, and when #

That seams wear first is as old as sewing and needs no explanation to anyone who has owned a shirt. That a pressed seam shines is a tailoring defect with a name and a standard remedy.

What this collection adds is the comparison of scales — that the step is two orders of magnitude larger than the depth at which cloth first touches anything, so the concentration is total rather than partial — and the identification of the shine with a calender, which makes it a computable quantity rather than a scorch.

Where the ladder goes next #

Nowhere further in this direction, for now. This is the last rung of the surface ladder: the height field that started as an answer to where does a cloth stop has been read for what touches, what bears, what wears, what reflects, what a coating fills and what two cloths do to each other, and the seam is where those readings meet a garment.

What remains unbuilt is stated in each essay’s own limits, and two of them recur. There is no model of a hair layer, and the hairs are what actually meet everything on every spun-yarn fabric. And the crimp ratio decides which of a cloth’s two systems is the one in contact, while remaining a convention rather than a measurement — which means the most basic question this ladder asks has an answer that nobody has weighed.