A seam must give what the knit gives

A knitted fabric reaches a hundred per cent extension at three newtons a metre. A sewing thread reaches three per cent before it breaks. Join two pieces of the first with the second and something has to give, and everybody in the trade knows what to do about it: sew more stitches per centimetre.

That rule has never had an arithmetic under it. It has one now, and it is a single line: the extension a seam can reach is twice the fabric’s thickness times the stitches per unit length.

Where the extension comes from #

Not from the thread stretching. A sewing thread’s own extension is a few per cent and it is used up immediately.

It comes from the thread being longer than the seam. A sewing thread does not run along the seam line; it goes through the fabric and back at every stitch, so the thread consumed per stitch is the pitch plus twice the fabric’s thickness. The thread available per unit of seam length is one plus twice the thickness over the pitch, and the extension available before the thread is taut is the second term alone.

Twice the thickness, times the stitches per unit length. Both quantities a maker sets directly.

The numbers #

A plain jersey of a 20 tex cotton is about a third of a millimetre thick — two yarn diameters, since a jersey is two layers of yarn where its loops cross.

At four stitches per centimetre a seam can reach twenty-seven per cent. At six, forty. At eight, fifty-three. At ten, sixty-seven.

Those are the numbers a knitwear maker works with, and the range they cover is exactly the range of stitch densities the trade uses. A rule of thumb that lands on the right numbers by accident would be a surprise; this one lands on them because it is the mechanism.

What the thread’s strength is for #

Almost nothing, on this account, and the margin is worth quoting to make the point.

A jersey at fifty per cent extension carries one and a half newtons per metre of seam. At seven and a half stitches per centimetre, one stitch carries two millinewtons. A 25 tex cotton sewing thread breaks at three and three quarter newtons.

That is a margin of nearly two thousand. The thread is not close to breaking under the fabric’s own load, and a seam that fails has failed for some other reason.

Which reasons those are #

Three, and only the first is about the thread.

The thread runs out of length. The seam reaches its geometric extension, the thread goes taut, and from there the load rises very steeply — the two-thousandfold margin is consumed in a fraction of a per cent, because the thread’s own modulus is four decades above the fabric’s. That is a snap rather than a stretch and it is the commonest knitted seam failure.

The thread cuts the fabric. A taut thread bearing on a loop is a concentrated load on a structure held by tens of millinewtons, and a stitch weakens what it passes through. The loops give way before the thread does.

The seam slips. A seam slips before it breaks in woven cloth, where the mechanism is threads sliding at the seam line. In a knit the same thing happens more readily, because the grip on a knitted thread is thirty times weaker.

All three are consequences of the seam having run out of extension, so the geometric line above is the thing to design to.

Why a stretch stitch is a different object #

An overlock or a chainstitch is not a lockstitch with more thread; it has a loop structure of its own that supplies extension by a second mechanism.

The thread in such a stitch forms loops that lie along the seam rather than only crossing it, and those loops can straighten out under extension in the same way a knitted loop does. So a stretch stitch has both terms — the through-thickness reserve counted above, and a loop reserve that can be several times larger.

That is why an overlocked seam reaches well past what its stitch density alone would allow, and why a lockstitch at any density cannot compete. The arithmetic here is the lower bound that applies to every stitch type; a stretch stitch’s own loop geometry adds to it and is not computed on this ladder.

What the picture cannot show #

A seam. Every figure here is a fabric, and the object being argued about is a thread passing through it at right angles to everything drawn.

Nor can any figure show the concentration. The load a stitch carries arrives at a few loops, and how it distributes among them decides whether the fabric fails before the thread — which is the question the stitch-weakening rung asked and which needs a model of a loop under a point load that nothing here supplies.

Where the fabric’s own load comes from #

The demand side of the arithmetic deserves the same treatment as the supply side, because it is the half that is usually assumed.

At a garment’s ordinary working extension a knit carries between one and three newtons per metre. A woven cloth of the same yarn at its working extension carries thousands. The three-decade difference is why a knitted seam has a two-thousandfold strength margin and a woven one does not, and it comes from the same place everything else on this ladder comes from — the loop having half its length spare.

So the seam problem is a consequence of the fabric’s softness rather than of anything about sewing, and it would go away entirely if knits were stiff.

The thread that never gets used #

There is a quantity worth naming because it is wasted in every knitted garment made.

A 25 tex sewing thread breaks at nearly four newtons and carries two millinewtons. Its whole tensile capacity is there to survive the moment the geometric reserve runs out — a fraction of a per cent of extension in which the load rises by three decades — and for nothing else.

That suggests an obvious substitution and the trade has made it: an elastic sewing thread, or a textured one with its own extension, moves the failure from a snap to a stretch and uses the strength margin to buy extension instead of holding a reserve. What this arithmetic adds is the size of the prize: the margin available to spend is a factor of two thousand, which is more room than any other seam decision has.

What the thickness is doing there #

It is the whole of the constant, so it is worth noticing which quantity it is.

A thicker fabric needs more thread per stitch, so a thicker fabric’s seam has more reserve at the same stitch density. That runs against intuition — a heavy fabric is usually the harder one to sew — and it is right: a heavy interlock sewn at six stitches per centimetre has more seam extension available than a fine jersey sewn at the same density, because each stitch buries more thread.

The catch is that the heavy fabric usually needs less extension, being less extensible itself. The two move together and the ratio between them is what a maker should actually be checking, and it is a ratio nobody computes.

Where the seam sits on the fabric’s curve #

A seam is not a point on the fabric; it is a line across it, and the fabric on either side is doing something the seam is not.

The consequence is familiar and rarely explained: a knitted seam puckers, or the fabric beside it ripples, because the two are on different parts of the same curve. A stitch density chosen for extension alone can produce a seam that survives every pull and looks wrong at rest.

Nothing here computes the pucker, which needs the fabric’s shear as well as its extension. What the arithmetic does say is that the effect is a consequence of the fabric being extensible rather than of anything the sewing did, so it cannot be sewn away — only designed around.

The specification this argues for #

Not a stitch density. A ratio: the seam’s available extension over the fabric’s working extension, with both computed rather than assumed.

The seam’s is twice the thickness times the stitch density, plus whatever the stitch type’s own loop supplies. The fabric’s is the extension the garment reaches in wear, which is a design decision. Asking for the first to exceed the second is one line of arithmetic and it replaces a rule of thumb quoted in stitches per centimetre with no reference to what is being sewn.

Where a woven seam differs #

Everything above inverts for a woven cloth, and the inversion is instructive.

A woven cloth reaches its own extension limit at a few per cent and at a large force, so a woven seam’s thread is genuinely loaded — seam strength is a real specification and threads do break. The extension mismatch that dominates a knitted seam barely exists, because the fabric is not going anywhere the thread cannot follow.

So the two trades’ seam specifications are answering different questions with the same words. A woven seam specification is about strength; a knitted one should be about extension, and it is usually written as a stitch density with no statement of what extension that buys.

Two rules that turn out to be one #

The trade has two separate pieces of advice for knitted seams, and the arithmetic shows they are the same advice.

Sew more stitches per centimetre, which buys extension by the line above.

Use a differential feed, which eases one ply relative to the other so that the seam is fed with a little slack in it. That buys extension too, by adding thread length along the seam rather than through it — a different term with the same units and the same effect.

Both are ways of putting more thread into a given length of seam, which is the only thing that helps. A third — a wider seam allowance — does nothing at all for extension, and is often reached for because it helps with slippage in woven cloth.

Knowing that the three advices divide into two that address extension and one that does not is the sort of thing an arithmetic buys that a rule of thumb does not.

Why the required density is nearly the same for every knit #

The specification this rung argues for is a ratio, and inverting it gives the number a sewing room would actually set. Requiring the seam’s reserve to cover the fabric’s working extension gives

stitches per unit length ≥ working extension ÷ (2 × thickness),

which for a jersey a third of a millimetre thick at fifty per cent working extension is 7.6 per centimetre — the density the trade uses, arrived at from two measurements and no rule of thumb.

The interesting part is what happens when the same inequality is asked of the other structures, because both of its terms move together.

A rib is about twice a jersey’s thickness, being two sets of loops leaning to opposite faces, and it works across about twice the extension, because its folding supplies a hundred per cent before the loops do anything. Both sides of the inequality double and the required density does not move.

An interlock is thicker again and less extensible, because its two ribs restrain each other, so the numerator falls while the denominator rises and the required density comes down.

So one stitch density serves the two structures a garment maker meets most, and the reason is not luck: the thing that makes a rib extensible is the same thing that makes it thick — the wales lying in two planes rather than one. Extension and thickness are two readings of the same doubling, and the ratio between them survives it.

That is a satisfying account of a rule of thumb that has always been quoted as one number. Six to eight stitches per centimetre is right for a jersey and right for a rib, for a reason, and it is generous for an interlock — which is exactly the structure a sewing room finds easy and never asks about.

Two cautions on the table. The interlock’s working extension is a design figure rather than a measurement, and moving it moves that row directly. And the thicknesses are two and four yarn diameters, which is a construction rather than a gauge reading — a heavily napped or a laminated fabric has more thread buried per stitch than its structure says and correspondingly more reserve.

What was known before #

The rule, in stitches per centimetre, for as long as knitted garments have been sewn. The requirement that a knitted seam use a stretch stitch, universally. The observation that knitted seams fail by snapping rather than by tearing.

What appears not to have been written down is the line connecting the stitch density to the extension, which is elementary once the thread’s path is followed and which nobody follows because the through-thickness excursion looks like a detail. It is not a detail: it is the entire reserve.

What the answer depends on #

Listing the inputs makes it clear how few there are, which is the strongest thing about the result.

The extension a seam can reach depends on the fabric’s thickness and the stitch density. Not on the sewing thread’s count, its fibre, its strength or its own extension. Not on the fabric’s fibre, count, loop length or tightness factor. Not on the seam allowance or the seam type, beyond whatever loop reserve the stitch adds on top.

That is a very short list for a quantity that decides most knitted garment failures, and it means most of what a specification usually controls is controlling something else. A thread chosen for strength is choosing a property with a two-thousandfold margin; a thread chosen for extension is choosing one that adds to the reserve and is the decision that matters.

The one input that is not on the list and should be is the stitch type’s own loop reserve, which is not computed here and which for an overlock is probably larger than the through-thickness term. That is the obvious gap and it is a geometry problem rather than a mechanics one.

What would test it #

Sew one fabric at five stitch densities and pull each seam to failure, recording the extension at which the load turns up sharply.

The prediction is a straight line through the origin with a slope of twice the fabric’s thickness — so five densities should give five points on one line, and the line’s slope should equal a thickness a gauge can measure independently. That is a strong test because it predicts a relationship rather than a value, and because both the slope and the intercept are predicted.

A lockstitch is the right stitch to test with, because a stretch stitch adds its own loop reserve and would offset the line by an amount this ladder does not compute.

The seam as a fabric of its own #

A last framing, because it makes the whole rung one sentence.

A lockstitch is the one seam type with no loop reserve: its thread crosses and locks, and the only spare length is the through-thickness excursion this rung counts. Every stretch stitch is a knitted structure sewn along a seam, and it works for exactly the reason a knit works.

So a knitted garment sewn with a chainstitch is a knit joined to a knit by a knit, and all three extend by loops reconfiguring at constant thread length. That is a satisfying way to end the ladder: the mechanism the whole thread has been about turns up one more time, at a scale nobody was looking at.

Where the ladder goes next #

Nowhere on this thread. The machinery that made these numbers available is spent, its limits are written down, and the obvious next piece — a rod with torsion in it, and a contact between neighbouring courses — is a larger object than this collection has built so far.