Textile Structure
how cloth is put together
Knits and other structures

The loop

A knit is one thread bent into loops, each drawn through the loop below. Nothing in it is straight, which is why it stretches in every direction while a woven cloth stretches only at an angle.
15 min read 6 figures Structure before fibreA mechanism, not a material

A woven cloth is two thread systems crossing. A knitted one is a single thread bent into loops, each loop drawn through the loop below it, and that difference in construction produces a difference in behaviour large enough that almost nothing said about one transfers to the other.

The knitted loopOne thread, bent into a course of loops, each of them drawn through the loop below. Nothing here is straight, which is why a knit extends in every direction while a woven cloth extends only on the bias.a course runs acrossa wale runs down5 wales × 4 coursesevery loop passes through the one below itweft knit
Fig. 1 A weft-knitted fabric: one thread, bent into a course of loops, each drawn through the loop below. Nothing here is straight, and nothing here is under tension along its length.

Wales and courses

The vocabulary first, because the two words are not warp and weft and using them interchangeably causes real confusion.

A course is a row of loops running across the fabric — in weft knitting, one traverse of the yarn. A wale is a column of loops running along it, each drawn through the one below.

They are not equivalent to warp and weft. In a woven cloth the warp and weft are different threads; in a weft-knitted one, a course is the thread and a wale is a chain of loops belonging to many different traverses of it. Pull a course out of a plain knit and it comes away as one continuous strand. Pull a wale and nothing comes away at all, because there is no such thread.

That asymmetry is the source of most of what a knit does, including how it fails.

Loop length is the variable

There is one number in knitting the way there is one number in weaving, and it is the length of yarn in a single loop.

Loop length — stitch length, in the trade — controls essentially everything. A long loop gives an open, soft, extensible, light fabric that snags easily and recovers poorly. A short one gives a firm, dense, stable fabric that resists distortion and feels harder.

It is the exact counterpart of float length in weaving, and for the same reason: it is the free length of yarn between constraints, and free length is compliance. Everything in this subject that behaves like a spring is a length of thread that has room to move.

Knitting controls it directly, which weaving cannot. A knitting machine’s cam setting decides how far the needle descends and therefore how much yarn each loop takes, and it is measured and controlled to a fraction of a millimetre because the fabric’s weight, width and handle all follow from it.

How a loop is actually made

The mechanism is worth a paragraph, because it explains why knitted fabric is made the way it is and why loop length is controllable in a way float length is not.

A latch needle holds the loop from the previous course in its hook. It rises, and the old loop slides down the shank and pushes the latch open. New yarn is laid into the hook. The needle descends; the old loop rides up, closes the latch over the new yarn, and slips off the end — drawing the new yarn through itself as it goes.

That is the whole operation, repeated a few thousand times a second on a modern machine. The height the needle descends to decides how much yarn is drawn into each loop, which is why loop length is set by a cam and measured to a hundredth of a millimetre.

The knitted loopOne thread, bent into a course of loops, each of them drawn through the loop below. Nothing here is straight, which is why a knit extends in every direction while a woven cloth extends only on the bias.a course runs acrossa wale runs down7 wales × 5 coursesevery loop passes through the one below itweft knit
Fig. 2 The result, over more wales and courses. Every loop was formed by exactly the operation above, and the fabric’s weight, width, handle and extensibility all follow from how far the needle went down.

The contrast with weaving is sharp. A weaver controls the sett by the reed and the beat-up, and the resulting crimp is a consequence rather than a setting — it emerges from the tension balance between two thread systems. A knitter sets the loop length directly and the fabric follows. That is why knitted fabric can be specified so precisely and why a change of yarn on a knitting machine usually needs no more than a cam adjustment.

Why a knit stretches

Take a knitted fabric and pull it sideways. It extends by half its length again, easily, and springs back.

Nothing stretched. The loops changed shape: what was tall and narrow became short and wide, and the yarn simply redistributed itself round a different-shaped path. The total length of yarn in each loop is exactly what it was.

That is the same class of explanation as crimp interchange in weaving and as the bias, and it is the recurring theme of this site: when a fabric appears to do something a material could not, a length has usually been rearranged rather than changed.

The difference is one of magnitude. A woven cloth’s crimp gives a few per cent in the thread directions. A knit’s loop geometry gives tens of per cent in every direction, because a loop can change its aspect ratio a long way before any yarn has to travel further.

Two extensions, two mechanisms

It is worth separating what happens when a knit is pulled in its two directions, because they are different.

Pulled along a wale — lengthways — the loops elongate: the heads and legs straighten, the loop becomes tall and narrow, and the fabric gets longer and narrower. Pulled along a course — crossways — the loops flatten: they become short and wide, and the fabric gets wider and shorter.

Both are the same yarn taking a different-shaped path, and both are bounded. The bound is reached when the loop has straightened as far as its own length permits, at which point further extension has to strain the yarn and the fabric becomes abruptly stiff. That knee is the same phenomenon as a woven cloth running out of crimp, at several times the extension.

And the two are coupled in the same way as crimp interchange: pull one direction and the other contracts, because the yarn taken by one has to come from the other. A knitted fabric held under lengthways tension is measurably narrower than the same fabric relaxed, which is why knitted goods are measured after relaxation and why the width of a roll of jersey is not a fixed property of the roll.

Why it needs no bias

A woven fabric is nearly inextensible along its threads and very extensible at forty-five degrees, so a garment that has to move with a body is either cut on the bias — expensive in cloth — or made with the ease built in by shaping.

A knit is extensible in every direction at once, so it needs neither. That single property is why knitted underwear replaced woven, why sportswear is knitted, and why a knitted garment can be made in far fewer pieces than a woven one: the fabric does the shaping that a woven cloth needs seams and darts for.

It is also why knitted fabric is harder to cut and sew accurately. A cloth that changes shape when handled will not stay where it is put, and the whole of knitted-garment construction is arranged around that.

Recovery, and where it comes from

Extension is one thing and coming back is another. A knit recovers because the yarn in a bent loop is elastically strained — a bent thread stores energy the way a bent spring does, and releasing it restores the loop’s shape.

That is a genuinely mechanical property, and it is where structure hands over to material. A knit in a fibre with poor bending recovery — cotton, notably — comes back imperfectly, which is why cotton knits bag at the elbows and knees. Wool recovers well because the fibre is naturally crimped and elastic; synthetics recover very well; and elastane is added precisely to supply recovery that the structure alone gives only partially.

So a knit’s extension is structural and its recovery is partly material. That is an unusual split in this subject, where structure usually dominates, and it is worth noticing that the exception exists.

Warp knitting is a different thing

The word “knitting” covers two families that share a loop and share almost nothing else.

Weft knitting is everything described so far: one thread traversing, loops formed in sequence along a course. It is what hand knitting is, what a sock machine does, and what most jersey is.

Warp knitting uses many threads at once, one per wale, each forming loops down its own column and moving sideways to interlock with its neighbours. Tricot and raschel are warp knits. They cannot be unravelled by pulling a thread, they are far more dimensionally stable, and they are made on completely different machinery.

The distinction matters because almost every generalisation about knits — that they ladder, that they curl, that they stretch enormously — is really a statement about weft knits. Warp knits share the loop and not the consequences.

The knit-purl distinction

One more piece of vocabulary that turns out to be structural rather than decorative.

A loop drawn through the one below can be pulled from either side. Pull it toward the observer and the loop’s legs face front — a knit stitch. Pull it away and the legs face back and the head faces front — a purl stitch. The same loop, the same yarn, the same length; only the direction it was drawn through.

Both faces exist in every fabric: a knit stitch on the front is a purl stitch seen from the back. What varies between fabrics is the arrangement, and the two simplest arrangements are the two everybody meets.

Stockinette — jersey — is every stitch knit. One face is all legs and smooth, the other all heads and bumpy. Garter alternates knit and purl courses, so both faces look the same.

That difference sounds cosmetic and is not. It decides whether the fabric lies flat, which is a mechanical consequence of a purely structural asymmetry, and it is the clearest available case of arrangement dominating everything else.

Why one curls and the other does notA knitted loop is not symmetric front to back. Worked every course the same way, the asymmetries add along the edges and the fabric rolls; worked alternately, consecutive courses point opposite ways and cancel.the dots mark which way each course facesstockinette — every course the same0 face changes · curlsgarter — alternate courses reversed5 face changes · lies flatface changes counted, not the curl itself
Fig. 3 The two arrangements, with a dot marking which way each course faces. Stockinette faces one way throughout and garter alternates, and that alone decides whether the fabric rolls up at its edges.

Rib and interlock

Two further arrangements are worth naming because they are everywhere.

Rib alternates knit and purl within a course rather than between courses — say two knit, two purl, repeating. The knit wales come forward and the purl wales go back, so the fabric pleats itself and its relaxed width is far less than its extended width. That gives enormous crosswise extension, which is why cuffs, collars and waistbands are ribbed.

Interlock is two rib fabrics knitted into one another, so that the fabric is identical on both faces and considerably more stable than either. It is heavier, does not curl, and does not ladder easily.

Both are the same loop in a different order. That is the pattern of the whole subject: a very small number of primitives, arranged, and the arrangement doing the work.

Against the woven case, point by point

Setting the two structures side by side is the fastest way to see what a loop buys and what it costs.

Extension. A woven cloth gives a few per cent along its threads and a great deal at forty-five degrees. A knit gives tens of per cent in every direction. The woven mechanism is crimp and shear; the knitted one is loop shape.

Stability. A woven cloth holds its dimensions well, because its threads are straight and gripped. A knit does not, because its loops can take many shapes at the same yarn length — which is why knitted garments need relaxing before they are cut and why they grow in wear.

Failure. A woven cloth frays at a cut edge and a break frees one thread. A knit ladders, and a break frees every loop above it. The reason is topological and it is the sharpest difference between them.

Air. A knit at rest holds far more air than a woven cloth of the same weight, because the loops are bulky and the structure is open. That is why knitwear insulates and why a woven windproof does not need to be heavy.

One break, two outcomesThe same single break in a knit and in a weave. In the knit nothing holds the loop above the break, so the failure climbs the wale; in the weave every other thread is still held by the threads crossing it, and one thread comes loose.a knit — the break runs up the wale3 loops freeda weave — the break stays put1 thread freedcounted from the structure, not from experience
Fig. 4 The difference that matters most in use. One break frees a whole wale in a knit, because nothing else holds those loops; one break frees one thread in a weave, because every other thread is held by the threads crossing it.

Cutting. A woven cloth can be cut in any direction and behaves predictably in each. A knit has to be cut with the wales running the intended way, because its extension is anisotropic and a garment cut the wrong way round hangs wrongly and grows in the wrong direction.

The list has a shape. Almost every advantage of the loop is compliance and almost every disadvantage is the same compliance seen from the other side, which is the trade that runs through the whole subject.

Gauge, and why it is not thread count

Knitting has its own density measure and it has the same problems as thread count, for the same reasons.

Gauge is stitches and courses per unit length — say twelve stitches and sixteen courses to the inch. It is measured on a relaxed swatch, and it is the number a pattern specifies because it determines the finished dimensions.

Two fabrics at the same gauge in different yarns are quite different cloths, exactly as with thread count. And the gauge of a knitted fabric depends on how it was relaxed: a swatch measured straight off the needles, after washing, and after washing and drying flat will give three different answers, sometimes differing by ten per cent. Knitting instructions insist on washing and blocking a tension swatch before measuring, and the insistence is well founded.

Why one curls and the other does notA knitted loop is not symmetric front to back. Worked every course the same way, the asymmetries add along the edges and the fabric rolls; worked alternately, consecutive courses point opposite ways and cancel.the dots mark which way each course facesstockinette — every course the same0 face changes · curlsgarter — alternate courses reversed5 face changes · lies flatface changes counted, not the curl itself
Fig. 5 Two arrangements at the same gauge, in the same yarn, at the same loop length. Everything a gauge measurement would report is identical, and the fabrics behave completely differently — one rolls at its edges and the other lies flat.

The underlying quantity, as ever, is not the count but the loop length, which industry measures directly by running a known number of stitches off a machine and weighing or measuring the yarn. That is the knitting equivalent of preferring cover to thread count, and it is more widely observed in knitting than the corresponding good practice is in weaving.

What this account leaves out

Three honest limits.

Loop shape here is schematic. A real relaxed loop takes a shape set by the balance of bending energies in the yarn, which is a genuine mechanics problem with a substantial literature — Peirce worked on it, as he did on woven geometry, and later models are considerably more sophisticated. The loops in these figures are drawn to be recognisable rather than solved.

Nothing here computes a force. How much a knit extends for a given pull, how much it recovers, and how it behaves after a hundred cycles are mechanical questions this site does not attempt.

Yarn matters more here than in weaving. Recovery is a fibre property, and a knit’s behaviour depends on it more than a woven cloth’s does. The usual claim that structure dominates is weaker in this field than elsewhere on this site.

Where the loop came from

Knitting is much younger than weaving and its history is unusually well marked by a single machine.

The oldest surviving true knitted fabrics are Egyptian, from around the eleventh century — considerably later than the earliest weaving, which predates writing. What came before, and is often mistaken for knitting in museum labels, is nålbinding: a single-needle looping technique that makes a similar-looking fabric and cannot be unravelled, because each stitch is worked through the yarn rather than merely through a loop.

The decisive event is 1589, when William Lee invented the stocking frame — a machine that formed a whole course of loops at once with a bank of hooked needles. It is the first machine to make a fabric rather than to assist a person making one, it was refused a patent by the Queen on the grounds that it would put hand knitters out of work, and its basic action is still what a modern knitting machine does.

The knitted loopOne thread, bent into a course of loops, each of them drawn through the loop below. Nothing here is straight, which is why a knit extends in every direction while a woven cloth extends only on the bias.a course runs acrossa wale runs down7 wales × 5 coursesevery loop passes through the one below itweft knit
Fig. 6 The structure the frame produced, and still produces. A bank of needles forming a course of loops at once, each drawn through the course below — the same operation Lee mechanised, at several thousand times the speed.

The latch needle came in 1847, which made the operation self-acting and is why knitting machinery is fast. Circular machines followed, then warp knitting, and the industry has been shaped since by the fact that a knitted fabric can be made faster than a woven one and shaped as it is made.

That last property is why knitting keeps taking territory from weaving. A machine that can shape a garment as it knits it removes the cutting and much of the sewing, and the fabric’s own compliance removes the darts that a woven cloth needs.

Where the ladder goes next

The clearest consequence of the loop’s asymmetry is why stockinette curls, which is a structural fact with a mechanical outcome.

The clearest difference from weaving is how the two fail: one break frees a whole wale in a knit and one thread in a weave, and the reason is topology.

What the pictures here cannot show. Every loop drawn here is a schematic curve, not a solved shape. And no figure on this page shows a knit under tension, which is the state in which almost all of the behaviour described here becomes visible.

Where this goes next

the dots mark which way each course facesstockinette — every course the same0 face changes · curlsgarter — alternate courses reversed5 face changes · lies flatface changes counted, not the curl itself Why stockinette curls A knitted loop is not symmetric front to back. Work every course the same way and the asymmetries add up along the edges; alternate them and they cancel. That is the whole difference between a fabric that rolls and one that lies flat. a knit — the break runs up the wale3 loops freeda weave — the break stays put1 thread freedcounted from the structure, not from experience Ravel, fray and run Cut a woven cloth and one thread comes loose. Break one loop in a knit and every loop above it follows. The two structures fail in opposite ways, and the reason is topology rather than strength. shear 30°bias +22.5%across -29.3%area 87%locks at 60°every segment checked against its own lengthno thread stretches The bias is a mechanism Cloth cut at forty-five degrees stretches by a third and springs back, while the threads in it stretch by nothing at all. Almost every explanation given for this is wrong, and the right one is not about elasticity.