The 6:40 a.m. roll
On the last Tuesday of January 2025, I had 1,600 metres of lining fabric opened on the inspection table in our receiving bay at 6:40 in the morning — earlier than usual, because the lot was already two weeks late and the cutting room was waiting on it.
I'm the quality and brand compliance manager at a mid-sized tailored-clothing manufacturer. I review somewhere around 300 fabric lots a year before they're released to production. In 2024 I rejected roughly 9% of first submissions, and most of those were for boring reasons: shade variation, width off-spec, a shrinkage figure that didn't match the tech pack.
This one wasn't boring.
The purchase order said 100% cupro, 58 g/m², plain weave, Asahi Kasei's Bemberg. We'd run Bemberg cupro linings on our jacket programme since before I joined. It drapes the way a jacket lining is supposed to drape, it breathes, it doesn't cling, and — this gets less credit than it deserves — it steams back into shape after a pressing error.
The lot in front of me looked right. Hand was close. Lustre was close. I signed the visual and sent it to the lab.
"Isn't cupro basically viscose?"
Two months earlier, in November 2024, one of our junior product developers asked me a question I gave a lazy answer to.
"What is viscose fabric, really? Isn't cupro just... viscose?"
I said something like, "They're both regenerated cellulose, but the process is different," and moved on. That was a mistake. Not because the answer's wrong — it's more or less right — but because "both regenerated cellulose" is exactly the kind of true-but-useless statement that teaches people to treat two fibres as interchangeable.
Here's the honest version.
Viscose fabric is made by dissolving cellulose pulp — usually wood — in sodium hydroxide and carbon disulfide to form cellulose xanthate, then extruding that into an acid bath to regenerate the fibre. The filament cross-section tends to be serrated or crenulated. It's soft, highly absorbent, and drapes well. It's also weaker wet than dry, which is why washing and drying behaviour has to be specified rather than assumed.
Cupro is also regenerated cellulose, but it's made by dissolving cotton linter in a cuprammonium solution and spinning it into a fine filament with a round cross-section. That round cross-section is a large part of why it feels the way it does — closer to silk in hand, very fine denier possible, good moisture take-up, minimal static.
Lyocell fabric is a third path. The cellulose is dissolved directly in N-methylmorpholine N-oxide and spun, with the solvent recovered in a loop. Its characteristics are genuinely different from viscose's: higher tenacity, especially when wet; a pronounced tendency to fibrillate, which can be suppressed or leaned into for a peach-skin surface; and noticeably different dyeing behaviour.
Same family. Three different animals. And the label on the roll only ever tells you one word.
We ran the qualification properly, which is what made it worse
After the junior's question, I did what I should've done a year earlier. We ran a three-way qualification for the spring 2026 lining programme: incumbent Bemberg cupro, a viscose candidate, and a lyocell candidate.
Bench set: ISO 1833-1 for quantitative composition, AATCC TM135 for dimensional change after home laundering, ASTM D5035 for strip tensile dry and wet, ISO 105-X12 for rubbing fastness, ISO 9237 for air permeability, and ISO 11092 on the sweating guarded hotplate for thermal and water-vapour resistance. Then a wear trial on twelve jackets.
The result annoyed me. The viscose candidate performed fairly well — better than I expected on dimensional stability across three wash cycles. The lyocell candidate was strong but stiff in the drape we needed; different fabric, different spec, not a drop-in. And the Bemberg reference did what it always does. Nothing dramatic. That's sort of the point of a reference sample.
I wrote the report, recommended staying with cupro on the core programme, and moved on. Which is why the January lot blindsided me.
Every number said ship it
The lab came back in four days.
Composition by ISO 1833-1: 100% cellulosic. No synthetic content, no cotton. Pass. Dimensional change, AATCC TM135: within our ±1.5% tolerance. Pass. Strip tensile, dry and wet: above our floor. Pass. Rubbing fastness: grade 4. Pass.
Here's the part I didn't understand until we went one level deeper. ISO 1833-1 tells you what a fibre is made of. It doesn't reliably tell you which regenerated cellulose you're holding, because cupro and viscose are the same polymer. Cellulose is cellulose. The method separates cellulose from polyester, from wool, from elastane — it doesn't separate cupro from viscose.
What separates them is morphology, and paperwork.
We sent a section to a fibre-identification lab for cross-section microscopy at 500×. The population wasn't uniform. Alongside round cupro-type filaments there was a visible share of serrated, crenulated filaments — the standard viscose profile. ATR-FTIR on the separated components came back consistent with that reading.
Then we asked for the traceability file. The mill could produce a spinning certificate for the yarn. It couldn't produce a Bemberg certificate of origin for the lot. Not a forged document — there simply wasn't one.
"The yarn is cupro." — the mill's written response, February 2025. Two weeks later they conceded they couldn't trace the lot back to a branded filament source.
What it actually cost
We rejected the lot. The supplier redid it at their cost, in about five weeks.
Take this with a grain of salt, but I'd put our internal damage somewhere in the $14,000–$18,000 range once you add air freight on the replacement, the pressing re-run on 340 jackets, and two days of my team's time on documentation. The supplier's number was higher.
What bothered me wasn't the money. It was that our system had no gate for this. We tested performance exhaustively and identity hardly at all. If the fabric had been 100% viscose and correctly labelled, I'd have caught the spec mismatch in a second. Because it was labelled cupro and behaved like cupro, nothing tripped.
From the outside, a lining roll is a lining roll and the paperwork is a formality. The reality is that once the fabric is woven, the paperwork is the only thing that distinguishes one regenerated cellulose from another.
What we changed
Three gates now, not one, and a pass on one doesn't cover a gap on another. Performance: the existing bench set — shrinkage, tensile, fastness, permeability, thermal resistance. Identity: cross-section microscopy on every new lot of a named-brand cellulosic, with ATR-FTIR if the morphology comes back mixed. Provenance: the brand-owner certificate arrives with the lot, not three emails later.
The third gate is the one people push back on. "Isn't that just bureaucracy?" It isn't, and the reason has changed in the last two years.
Five years ago, a fibre name on a purchase order was reasonable shorthand. You bought cupro, you got something that behaved like cupro, everyone was broadly satisfied. That's no longer a safe assumption — substituted and blended cellulosics are more common in the market than they were, and the regulatory floor is moving. Regulation (EU) 2024/1781, the Ecodesign for Sustainable Products Regulation, entered into force on 18 July 2024, and textiles sit in the first group of product-specific rules being drafted under it. I'm not going to pretend I know what the final data fields look like; nobody does yet. But the direction is unmistakable. A digital product passport will ask what your fibre is, and "regenerated cellulose" won't be an answer.
What was good practice in 2020 — name on the PO, certificate on request — doesn't hold up in 2026. The fundamentals of a good spec haven't changed. The evidence standard attached to it has.
The question I ask now
I used to ask suppliers, "Is this 100% cupro?" Nearly everyone says yes.
The question I ask now is: "Which cellulose, and can you show me the chain of custody for this lot number?"
That's the outsider blindspot in this category. Buyers focus on the fibre name on the invoice and completely miss that the fibre name is the weakest link in the entire specification. The name is a claim. The certificate is evidence. They aren't the same document, and only one of them survives a microscope.
We kept Bemberg cupro on the core programme — not out of habit, we re-tested it against both alternatives and it won on the hand-and-recovery combination our jackets need. But I'd probably have kept it anyway. What changed is that I now make the supplier prove which cellulose I'm buying, every lot, before it goes anywhere near a cutting table.
Which is, honestly, what I should've been doing since 2022.