BRENNEN

Industries / Consumer goods / Textiles and apparel

Sorting and recycling

Sorting and recycling collects used clothes and factory scraps, sorts them by fiber and color, removes zips and buttons, and turns the cloth back into fiber, either by tearing it apart (mechanical recycling) or by dissolving it into pulp or plastic (chemical recycling).

7

Milestones achieved, from 1813 to 2025.

3

To watch for: announced, not yet confirmed outside.

3

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1813

    Rags become new cloth

    Mills in Batley, Yorkshire, began tearing old woolen rags back into fiber, called shoddy, and spinning it into new cloth. Samuel Jubb's 1860 history dates the start to 1813 and says the credit probably belongs to Benjamin Law of Batley, though nobody is sure. By 1860 Batley had more than 80 rag firms, and dealers sorted rags into more than 20 grades by color and quality.

    LessonCareful sorting by color and quality is what made recycled fiber worth buying.

    archive.org

    Credit

    probably Benjamin Law

  2. 2007

    France makes brands pay

    France made every company that sells new clothes, household linen and shoes pay for collecting and recycling them, a rule called extended producer responsibility (EPR). The scheme is run by Eco TLC, now called Refashion. A 2018 paper by Bukhari and colleagues calls it the first legal framework for textile waste based on EPR.

    LessonMaking producers pay raised collection, which tripled from 65,000 tonnes in 2006 to about 210,000 tonnes in 2016, by the same paper's figures.

    senat.frcirculareconomy.europa.eu

    Credit

    the French Parliament, in article 69 of the 2007 finance law

  3. 2017

    Measuring the recycling gap

    The Ellen MacArthur Foundation estimated that "less than 1% of material used to produce clothing is recycled into new clothing." Textile Exchange's 2025 report found that pre- and post-consumer textile recycling supplied less than 1% of the world's fiber in 2024.

    LessonMost recycled fiber does not come from old clothes; the same Textile Exchange report puts all recycled fibers at 7.6% of the market.

    content.ellenmacarthurfoundation.orgtransition-pathways.europa.eu

    Credit

    Ellen MacArthur Foundation; Textile Exchange

  4. 2020

    Machines sort by fiber

    Sysav's SIPTex plant in Malmö, Sweden, began sorting used textiles automatically in November 2020, with a stated capacity of 24,000 tonnes a year. It shines near-infrared (NIR) light on each item; each fiber reflects the light differently. Sysav called it the world's first large-scale plant for automatic textile sorting.

    LessonIn December 2025 Sysav's board decided to sell the plant because the market for sorted textiles grew more slowly than expected; its chief executive said, in Swedish, that Sysav was too early.

    sysav.mynewsdesk.commynewsdesk.com

    Credit

    Sysav

  5. 2020

    Labels proved unreliable

    Circle Economy, working for the Dutch government, scanned more than 10,000 garments with the Fibersort NIR machine and found that 41% of care labels did not match the measured fiber. In June 2026 an EU testing campaign found wrong labels on 49 of 132 garments (37%), and on 64% of those labeled as a blend of natural and artificial fibers.

    LessonSorters have to measure fiber, not trust labels.

    fashionunited.uksingle-market-economy.ec.europa.eu

    Credit

    Circle Economy; EU market surveillance authorities with the European Commission

  6. 2024

    A recycler fails, then restarts

    Renewcell turned worn cotton into a pulp called Circulose, used to make new fiber. It filed for bankruptcy in February 2024, after demand for its pulp "was slower than expected," as Waste Dive reported. In June 2024 the investment firm Altor bought its assets and renamed the business Circulose.

    LessonRecycled fiber needs buyers who commit in advance; Circulose's restart rests on volume commitments from brands, not one-off sales.

    inderes.sewastedive.comfashionunited.uk

    Credit

    Renewcell; Altor

  7. 2025

    EU makes textile collection law

    From 1 January 2025, EU countries must collect textiles separately from other waste. Directive (EU) 2025/1892, published on 26 September 2025, requires every EU country to set up textile EPR schemes by 17 April 2028.

    eur-lex.europa.eueur-lex.europa.eu

    Credit

    the European Parliament and the Council of the EU

Watch for

  1. Circulose plant back running

    Circulose says its Ortviken plant in Sundsvall, Sweden, is back in operation with commitments from 16 brands. In February 2026 it was still filling orders from pulp that Renewcell made before the bankruptcy.

    Achieved whena fiber maker or brand outside Circulose confirms it received pulp made at Ortviken after the restart, or audited figures give tonnes produced.

    textileworld.com

    Claimed by

    Circulose, November 2026

  2. Polyester from old clothes

    Syre of Sweden and JEPLAN of Japan say they are "targeting multi-ton volumes of textile-to-textile generated polyester in late 2026" for spinning trials and brand samples.

    Achieved whena spinner or brand outside both companies reports yarn made from it, and an independent certifier confirms the raw material was textile waste.

    textileworld.com

    Claimed by

    Syre and JEPLAN, April 2026

  3. Lasers cut off the zips

    Rochester Institute of Technology (RIT) built a line that uses infrared cameras and AI to find buttons, zips and fiber type, then cuts the hard parts off with a laser, processing each item in about 10 seconds. It was in pilot tests, funded by a $1.3 million grant from the REMADE Institute.

    Achieved whena recycler outside RIT runs it on its own used clothing and publishes the share of trims removed and the fabric lost.

    resource-recycling.com

    Claimed by

    RIT, July 2025

Open

  1. 01

    Lab-checked spectra of worn clothes

    A free, public set of NIR readings (spectra) from real worn garments, including blends, where each garment's fiber content is measured by an accredited lab instead of read from its label, with a scoring script.

    The brief
    Credit · unclaimed
  2. 02

    Flag stretch fiber in jeans

    A quick yes-or-no test with a handheld NIR scanner that says whether a worn cotton garment contains elastane (stretch fiber), which jams shredders and spoils chemical recycling.

    The brief
    Credit · unclaimed
  3. 03

    Public test set for trims

    Public photos of worn garments, both sides and inside out, with every hard trim (zip, button, rivet, snap) outlined, plus a scoring script, so trim-removal machines can be compared on the same garments.

    The brief
    Credit · unclaimed

Open 01 · the brief

Lab-checked spectra of worn clothes

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A free, public set of NIR readings (spectra) from real worn garments, including blends, where each garment's fiber content is measured by an accredited lab instead of read from its label, with a scoring script.

Why it is still open

The US National Institute of Standards and Technology (NIST) published its NIR-SORT dataset in May 2026, but took the fiber content of its second-hand garments from care labels and does "not recommend using" them for training until it cross-checks them. VTT's OpenTextile-NIR set (February 2026) has 71 factory-waste samples whose content was not verified by chemical tests.

nature.compmc.ncbi.nlm.nih.gov

Built with what exists

handheld NIR scanners like those NIST used (Matoha, Sortile); unsellable charity-shop garments; an accredited textile testing lab for fiber content, with no chemistry done at home; Python and scikit-learn; Zenodo for hosting.

  1. First step

    Collect 30 unsellable garments, scan five spots on each side, and get quotes from two accredited labs.

  2. Done looks like

    At least 300 worn garments, at least 30% of them blends, each with a lab report and readings from two instruments, published under an open license with a baseline accuracy on held-out garments that an outside team reproduces within 2 percentage points.

Open 02 · the brief

Flag stretch fiber in jeans

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A quick yes-or-no test with a handheld NIR scanner that says whether a worn cotton garment contains elastane (stretch fiber), which jams shredders and spoils chemical recycling.

Why it is still open

Aalto University and VTT detected 2–6% elastane only in 201 new, shop-bought fabrics in the lab. They name the variety of post-consumer textiles as the challenge for industrial use, and describe current checks as "invasive and time-consuming."

aaltodoc.aalto.fi

Built with what exists

a handheld NIR scanner; one-class models such as the SIMCA method Aalto used, or a small neural network; lab reference tests shared with the dataset above.

  1. First step

    Collect 40 worn jeans, 20 labeled stretch and 20 labeled 100% cotton, scan them, and lab-test 10.

  2. Done looks like

    On at least 200 worn, cotton-rich garments with lab-verified content, it flags at least 95% of those with 2% or more elastane and wrongly flags no more than 10% of pure cotton ones, scored by someone outside the team on held-out garments.

Open 03 · the brief

Public test set for trims

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

Public photos of worn garments, both sides and inside out, with every hard trim (zip, button, rivet, snap) outlined, plus a scoring script, so trim-removal machines can be compared on the same garments.

Why it is still open

Kingston University's 2025 paper on finding trims used 80 images and notes "the scarcity of large, publicly available datasets." Trim removal is still mostly done by hand; the Hong Kong Research Institute of Textiles and Apparel (HKRITA) says the "current method to remove trims and accessories is labour-intensive."

arxiv.orghkrita.com

Built with what exists

a phone camera and a light box; open-vocabulary detection and segmentation models such as Grounding DINO and SAM as a no-training baseline, as Kingston used; optionally an SO-101 arm with LeRobot to flip garments the same way each time.

  1. First step

    Photograph 50 unsellable garments, outline the trims, and run the baseline.

  2. Done looks like

    At least 1,000 garments and 5,000 outlined trims under a Creative Commons license, with a baseline recall (the share of real trims found) on 200 held-out garments that an outside team reproduces.