Stretch is not one thing
The Old English “streccan” means to extend, to lay straight; its past participle survives as the word straight. Stretch began as a word about going out, not coming back. The industry still uses it that way — and that is the problem. In a fabric, going out is the easy half. Everything that matters happens on the return.
Four numbers wearing one word
When a spec sheet says stretch, it is compressing four separate properties into one syllable.
Elongation is how far the fabric extends under a defined load.
Recovery is how much of that extension returns when the load is released.
Growth is what never returns — the permanent set left after prolonged extension.
Power is how hard the fabric pulls back on the way home: the difference between a waistband that holds and one that digs.
Brands buy on the first number. Garments fail on the third. Fit is decided by the fourth. The second is the only one anyone remembers to ask about, and it is usually fine.
The tests exist and are specific. ASTM D2594 covers knits under low tension: stretch under a known load, growth measured after an extension is applied, held, and removed. D3107 covers wovens made from stretch yarns; D4964 and D6614 cover elastic fabrics on constant-rate machines. In practice: specimens two hundred by fifty millimetres, reference lines a hundred millimetres apart, conditioned at 21°C and 65% humidity, loaded at 0.5 kilograms for low-power fabric or 2.3 for higher, sixty seconds of recovery, course and wale measured independently. A four-way claim requires each direction to pass on its own.
And then the fact that turns this from a summary into an argument. D2594 states, in its own text, that it is not recommended for acceptance testing of commercial shipments, because agreement between laboratories is known to be poor. Brands write stretch minimums into purchase orders and reject bulk against a method whose own authors tell you not to use it that way. The rejection letter and the standard disagree with each other, and the standard is right.
Two ways to make it, and a third nobody counts
Elastane is the answer everyone knows. A polyurethane fibre capable of extending five hundred percent, present at two to five percent of a garment, invisible on the care label. It supplies what nothing else supplies: power, and near-total recovery. Roughly 1.4 million tons of it were produced in 2025, most of it destined for apparel, and the volume is still climbing.
Bicomponent yarn is the answer the industry had before elastane got cheap. Two polyesters — PET and PTT — extruded side by side; their different melting points force differential shrinkage in heat treatment, setting a permanent helical crimp into the filament. The yarn is a spring by geometry, not by chemistry. One published comparison is worth carrying in your head: in denim, a bicomponent construction gave 22.0 percent stretch with 2.3 percent growth, against an elastane control at 25.3 percent stretch and 3.0 percent growth. Less extension. Better return. That pair of numbers is most of this essay.
The third source is the one that never appears on a fibre content label: construction. In wovens, high-twist yarns coil, and the weave geometry does the rest — ten to eighteen percent of stretch with no elastic fibre at all, and no degradation under repeated washing, bleach, or industrial laundry heat. In knits, gauge, structure, float length, and heat-set temperature move the stretch figure as much as fibre content does. The same yarn on the same machine, finished three degrees apart, is two different fabrics. We say this from the finishing floor, not from a textbook.
Why the wrong number gets specified
Growth is missing from most spec sheets because it cannot be felt in a showroom. It reveals itself in the third month of wear, in a legging that fits differently in March than it did in January, and by then the fabric decision is two seasons old.
Elastane percentage is treated as a performance figure when it is only an input. Three percent in a loose jersey and three percent in a dense interlock are not comparable products; the number says what went in, not what comes back.
Loads go unstated. Half a kilogram and 2.3 kilograms produce different elongation figures on the same cloth, so a stretch percentage without its load is not comparable across mills. It is not really a number.
And power — the property that decides whether compression compresses — is almost never specified at all, because it requires deciding what the garment is for, and that decision is easier to defer to the fabric.
The reversal now underway
Elastane became universal for one reason and is being reconsidered for a completely different one.
It won on cost and hand. It is losing on end-of-life. At two to five percent of a garment, elastane disrupts both mechanical and chemical recycling: it will not pass the shredders that open textiles for recovery, clogging and clumping in the machinery, and fabrics above roughly five percent are routinely downcycled into insulation rather than remade into garments. The EU has required separate textile collection since January 2025, and its recyclability rules increasingly tie fibre choice to compliance and cost. Even the optimistic projections put fibre-to-fibre recycling at eighteen to twenty-six percent of textile waste by 2030. The fibre that made stretch cheap is now the fibre that makes circularity expensive.
The industry's response has forked. One branch keeps the chemistry and changes the feedstock — bio-based elastane, with capacity investments in Vietnam scaling toward two hundred thousand tons a year. The other branch removes the chemistry entirely and returns to mechanical stretch: bicomponent yarns and high-twist constructions that survive the wash, survive the shredder, and hold their growth number. Mechanical stretch is not returning because it stretches better. It doesn't. It is returning because the questions changed.
What we will publish next
One base construction, knitted three ways on our own machines: with elastane, with bicomponent yarn, with construction alone. Elongation, recovery, growth, and power measured identically across all three, course and wale separately, loads stated. [Studio data to follow.] Then the same greige heat-set at two temperatures, to show that finishing moves the stretch figure as much as fibre content does — which is why the fabrics in our library are named for a process and a number, not a benefit.
Almost nobody can run this comparison, because almost nobody can make all three fabrics on the same machine with the same yarn count. We can. That is the only reason this journal exists.
Measured values in this journal carry their test standard. Assessments made by hand are labelled as studio assessments. Claims we cannot verify are omitted.
