The Verification Venue · a thing everyone half-knows

The Give Was Never in the Yarn

A knitted jumper stretches. A woven shirt, pulled along the grain, barely does — and it can be spun from the same yarn. So the stretch is not in the thread. It's in the shape the thread is held in: loops that rearrange, or a grid that jams.

Here is the thing worth doing in ten seconds. Pull a mesh of interlocking loops and the loops rotate and flatten — the fabric widens by tens of percent while the yarn itself never lengthens at all. Then pull a woven grid on the grain: it locks almost at once. Turn it forty-five degrees and pull on the bias, and the same "rigid" grid shears open like a garden trellis. Everything below is computed live from the real geometry — the yarn's length is conserved throughout, so the number that never moves is the one everyone blames.

Press a fabric, then pull it ↓

What you just did is exactly the three-ingredient model Poincloux, Adda-Bedia and Lechenault derived from first principles in Physical Review X (2018): a knit's shape and stiffness fall out of the yarn's bending energy, the conservation of its total length, and the topological constraints of the stitches. Stretching the fabric bends the yarn into new loop shapes; it does not pull the molecules of the fibre apart. That is why a knit "can produce loose yet extremely stretchable fabrics with almost vanishing rigidity, a desirable property exhibited by hardly any bulk material."

The woven story is the honest twin. A plain weave is close to a pin-jointed net of inextensible threads. Along the grain there is nowhere for length to come from, so it gives only a few percent — the warp de-crimps a little, the weft crimps up to match, and then the threads jam. But rotate the pull to the bias and the square cells shear into rhombi. No thread stretches; the crossing angle just changes. The extension is pure geometry:

So the knit widthwise (~39% here, at zero yarn strain) and the woven on the bias (~37%, at zero yarn strain) end up in the same league — both give freely — while the woven on the grain (~4%) is the outlier. The lesson isn't "knits stretch and wovens don't." It's that give is a property of the architecture, and the stiff-feeling woven does the very same trick, only diagonally.

Now drop a stitch

A knit's superpower and its curse are the same fact: every loop is held only by its neighbours. Cut one thread or let one loop off the needle and the loop below has nothing to hold it — it unmeshes, and the one below that, and a ladder runs down the wale (Americans call it a run; the British call it a ladder — same thing, in your tights). In 2026, Faulconnier, Michel, Adda-Bedia, Crassous and Steinberger showed this is a topology-induced failure with a clean control knob: a force threshold set by the stitch's natural curvature governs both when a ladder starts and when it stops. Slide the tension and watch the regime flip.

Contrast the woven again: a woven doesn't ravel, it frays — discrete straight threads slide out of independent crossings, one at a time, with no cascade. The knit's fragility is the price of its give: the loops that let it rearrange are the loops that can unmesh. And the resistance you feel pulling a knit is never zero — it's the yarn's bending plus the friction where loops press on each other at their crossings, which is why the real force-extension curve is soft, non-linear and hysteretic, not a limp string that goes slack and then snaps taut.

A quick glossary, because the words are traps. A wale is a vertical column of loops; a course is a horizontal row. "Course-direction stretch" means stretching along a course — i.e. widthwise. A knit and a purl are the same loop seen from opposite faces (which is why a stockinette curls). And "stretch" wovens in your jeans cheat: they add an elastomeric fibre (elastane / spandex) that itself stretches — a different mechanism entirely. Everything on this page is geometry from a single ordinary yarn, no elastane.