BRENNEN

Industries / Food, farming and forestry / Agriculture and farming / Cotton ginning

Cottonseed

Handling the seed that the gin strips of its fiber, and turning it into oil, meal and hulls or keeping it safe to feed.

6

Milestones achieved, from 1834 to 2018.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1834

    A mill for waste seed

    Through most of the 19th century, seed left at the gin was "considered a waste product and a nuisance": piles of it smoldered and stank. The Texas State Historical Association (TSHA) says the first commercially successful cottonseed-oil mill in the United States was built in Natchez, Mississippi, in 1834. By TSHA's figures, crushing splits the seed by weight into about 16% oil, 45% meal (the ground, de-oiled kernel, sold as high-protein feed) and over 25% hulls (the hard seed coat).

    tshaonline.org

    Credit

    the Natchez mill (TSHA names no builder)

  2. 1884

    A trust buys the crushers

    Mill owners' syndicates combined about 1884 into the American Cotton Oil Trust, which by the fall of 1886 controlled 88% of US crushing capacity, according to TSHA. Louisiana sued to dissolve it, and in 1889 it was reorganized as the American Cotton Oil Company. Mills kept multiplying, to 882 across the country in 1915; then overcapacity, soybeans and solvent extraction (dissolving the oil out instead of pressing it) cut Texas from 233 mills that year to 19 in 1983.

    Lessoncrushing capacity was built faster than the seed supply grew, and most of it closed.

    tshaonline.org

    Credit

    American Cotton Oil Trust; State of Louisiana

  3. 1911

    Shortening from cottonseed alone

    Procter & Gamble launched Crisco in 1911, which historian Helen Zoe Veit calls "the world's first solid shortening made entirely from a once-liquid plant oil." It hardened cottonseed oil by hydrogenation, adding hydrogen to the oil, instead of mixing the oil with beef fat as earlier brands did. That built on chemist David Wesson's bleaching and deodorizing of cottonseed oil in the late 19th century, which made it clear and tasteless. Crisco has since replaced cottonseed oil with palm, soy and canola oils.

    Lessona by-product sold under someone else's brand can be swapped out for another oil.

    smithsonianmag.com

    Credit

    Procter & Gamble; David Wesson

  4. 1950s

    Glandless cotton fails

    Gossypol, a compound held in dark glands in the seed, is toxic to people and to animals with one stomach, so most cottonseed protein goes to cattle. Breeders made "glandless" cotton with no gossypol in the 1950s. Texas A&M researchers wrote in 2006 that these varieties were "commercially unviable," because plants without gossypol anywhere were more easily attacked by insects.

    Lessonthe compound that makes the seed toxic also protects the plant.

    doi.org

    Credit

    1950s cotton breeders, as reported by G. Sunilkumar, Keerti Rathore and colleagues

  5. 2003

    A harmless mold protects seed

    About 1970, Arizona dairies dumped milk that carried aflatoxin, a poison made by the mold Aspergillus flavus, traced back to cottonseed fed to the cows; Farm Progress reports that federal rules bar cottonseed with more than 20 parts per billion (ppb) of aflatoxin from dairy feed. Peter Cotty of USDA's Agricultural Research Service found a strain of the same mold that makes no aflatoxin and crowds out the strains that do; AF36 was his 36th isolate. EPA conditionally registered it for cotton in Arizona and Texas on 24 June 2003. In four years of trials near Yuma, the Arizona Cotton Research and Protection Council (ACRPC) says, toxin-producing strains fell from 82% to 4% of the mold population.

    www3.epa.govfarmprogress.com

    Credit

    Peter Cotty, USDA ARS; ACRPC, the registrant

  6. 2018

    Seed without the poison

    Keerti Rathore's team at Texas A&M AgriLife Research used RNA interference, a way to silence one gene, to stop gossypol being made in the seed while leaves and flowers keep their normal, protective levels. On 17 October 2018 USDA's Animal and Plant Health Inspection Service (APHIS) stopped regulating the line, TAM66274, finding it "unlikely to pose a plant pest risk"; Texas A&M says FDA approved it in 2019. In 2025 the trait was available only for noncommercial use, and Rathore described it as a largely untapped opportunity for seed companies.

    Lessonapproval does not put a trait into the varieties farmers plant.

    govinfo.govagrilifetoday.tamu.edu

    Credit

    Keerti Rathore and Texas A&M AgriLife Research, funded by Cotton Incorporated

Watch for

  1. Low-gossypol seed for Uzbekistan

    Texas A&M Innovation agreed with Uzbekistan's Center of Genomics and Bioinformatics to breed TAM66274 into cotton varieties suited to Uzbekistan, with the seed as a possible food protein; Texas A&M says Uzbekistan is the first country to formalize such a partnership with it.

    Achieved whenan Uzbek variety carrying the trait is registered and grown on farms, and a lab outside Texas A&M reports low gossypol in its harvested seed.

    agrilifetoday.tamu.edu

    Claimed by

    Texas A&M AgriLife, July 2025

Open

  1. 01

    A camera screen for aflatoxin

    A light-tight box with 365 nm ultraviolet lamps and a fixed camera that photographs a measured sample of ginned seed, and a model that counts the bright greenish-yellow glow on the fuzz and predicts whether the lot is above 20 ppb aflatoxin, the dairy-feed limit.

    The brief
    Credit · unclaimed
  2. 02

    Pocket check for gossypol

    An open calibration that lets a pocket near-infrared (NIR) scanner, which measures how a sample reflects invisible light, estimate gossypol in whole cottonseed and in cottonseed meal in under a minute, so a feed mill can tell whether a lot suits poultry or fish feed and a gin can confirm that low-gossypol seed has stayed separate.

    The brief
    Credit · unclaimed

Open 01 · the brief

A camera screen for aflatoxin

Credit · unclaimed

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

What to build

A light-tight box with 365 nm ultraviolet lamps and a fixed camera that photographs a measured sample of ginned seed, and a model that counts the bright greenish-yellow glow on the fuzz and predicts whether the lot is above 20 ppb aflatoxin, the dairy-feed limit.

Why it is still open

The glow under ultraviolet light has been studied on cotton since the 1970s, but by eye, lock by lock. In 1978 Louise Lee and Alva Cucullu of USDA's Southern Regional Research Center hand-ginned 51 seeds from glowing locks and found that 88% of the seeds that would be thrown out with those locks held no aflatoxin, so the glow alone is a weak test. Machine vision has been built to sort broken cottonseed for planting, and camera and hyperspectral screens have been tested on peanuts, figs and corn, but I found no study or public image set that scores the glow on whole lots of commercial cottonseed against lab aflatoxin values.

api.crossref.orgdoi.org

Built with what exists

365 nm UV LED lamps; a phone or Raspberry Pi HQ camera in a blackout box; Label Studio to mark glowing spots; a small detection model such as Ultralytics YOLO, trained on a laptop; aflatoxin values from a commercial feed lab. Handle samples with gloves, a dust mask and UV-blocking glasses, and feed none of them.

  1. First step

    Ask an Arizona or Texas gin and its feed lab for 30 retained seed samples with their aflatoxin results, and photograph them under UV.

  2. Done looks like

    On 300 seed lots with lab aflatoxin values, held out from training, the screen flags at least 90% of lots above 20 ppb while passing at least 70% of lots below it; images, labels and lab values are published.

Open 02 · the brief

Pocket check for gossypol

Credit · unclaimed

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

What to build

An open calibration that lets a pocket near-infrared (NIR) scanner, which measures how a sample reflects invisible light, estimate gossypol in whole cottonseed and in cottonseed meal in under a minute, so a feed mill can tell whether a lot suits poultry or fish feed and a gin can confirm that low-gossypol seed has stayed separate.

Why it is still open

Benchtop NIR instruments already do this in the lab. Lordelo and colleagues predicted free gossypol in cottonseed meal in 2008, with r² of 0.73 on samples from around the world and 0.92 on US samples alone; Li and colleagues did the same for intact seeds from breeding plots, published in the Journal of Cotton Research in 2021. I found no published, open calibration for a low-cost handheld scanner on commercial seed and meal.

agris.fao.orgdoi.org

Built with what exists

A handheld NIR scanner such as Si-Ware's NeoSpectra; reference gossypol values by HPLC from a university or commercial lab; partial least squares regression in Python with scikit-learn; Zenodo to publish spectra and lab values.

  1. First step

    Buy 20 bags of whole cottonseed and meal from feed stores, scan each one, and send them to a lab for gossypol.

  2. Done looks like

    On 200 lots of seed and meal from at least three countries, held out from calibration, predictions fall within 0.1 percentage point of the lab's gossypol value for at least 80% of lots.