In 2015, a reconstructed medieval eye remedy cut recoverable Staphylococcus aureus in laboratory biofilms by four to five orders of magnitude. In a second experiment, infected wound tissue removed from mice and soaked in the mixture for four hours yielded about 90 percent fewer viable MRSA cells.
This was not a treatment trial in people, or even a four-hour treatment of living mice. The comparison took place in excised tissue after the animals had been euthanized, a boundary that matters when the result is set beside a modern antibiotic.
This is research reporting, not treatment advice. The mixture should not be applied to an eye or wound.
The recipe came from a tenth-century medical book
The starting point was British Library Royal MS 12 D XVII, a parchment manuscript known as Bald’s Leechbook. Its contents include medical remedies, diagnoses, charms and prognostic texts. The first two books were compiled for a person named Bald, according to a colophon, although the surviving manuscript is not a laboratory notebook in any modern sense.
One instruction describes a salve for a “wen,” or lump, in the eye. The condition was probably a stye, an infection of an eyelash follicle commonly caused by S. aureus. The healer was told to pound equal amounts of garlic and cropleac, add equal amounts of wine and oxgall, place the mixture in a brass vessel for nine nights, strain it and apply it with a feather.
Cropleac does not map neatly onto one modern plant name. Freya Harrison and colleagues therefore reconstructed two versions, one with onion and one with leek. Oxgall supplied bovine bile. Instead of an ancient bowl, the team used glass bottles containing sterile squares of brass sheet to imitate contact with a copper-alloy vessel.
They also wrapped the bottles to block light and stored them at 4 degrees Celsius for nine days. The cool temperature was a practical attempt to avoid the large daily swings of a heated laboratory, not a claim that an Anglo-Saxon healer owned refrigeration.
Nine days produced the strongest biofilm killing
The peer-reviewed paper, led by Harrison and published in mBio, began with free-floating bacterial cultures. Both complete salves prevented visible growth, and transferred samples did not regrow in fresh medium. That indicated killing rather than a temporary pause in growth under those conditions.
The more demanding experiment used one-day-old S. aureus biofilms in collagen-based synthetic wounds. Biofilms are dense communities of microbes held within a self-produced matrix. They can behave very differently from isolated cells in a flask and are often harder for antimicrobial drugs to penetrate.
Freshly prepared onion and leek salves reduced viable cells by about two orders of magnitude. The same recipes, after the prescribed nine-day wait, produced reductions of four to five orders of magnitude relative to water controls. A four-log drop corresponds to 99.99 percent fewer recoverable cells, while a five-log drop corresponds to 99.999 percent fewer.
That supports the title’s “more than 99.99 percent” description for the strongest results, but it is not a cure rate. It is a comparison of colony-forming units recovered from a small laboratory model after 24 hours of exposure.
This is one study, not settled consensus. The timing test used five biofilm cultures per treatment, and the researchers also observed some batch-to-batch variation.
The four-hour mouse-wound test used excised tissue
For the MRSA experiment, six adult female Swiss-Webster mice were given wounds infected with the strain Mu50. Four days later, the mice were euthanized. Researchers removed the wound tissue, divided it into pieces and submerged those pieces in saline, the onion salve, the leek salve or vancomycin at 100 micrograms per milliliter.
After four hours, both salve variants produced statistically significant drops of approximately tenfold in viable MRSA recovered per gram of tissue. A tenfold reduction leaves about one-tenth as many cells, so describing the result as “about 90 percent” is reasonable.
Vancomycin did not produce a statistically significant reduction during the same brief exposure. The paper reports a P value of 0.491 for that comparison, well above the conventional 0.05 threshold.
This does not establish that the salve works better than vancomycin as a treatment. The antibiotic arm drew one tissue piece from only three mice, the exposure lasted four hours, and the test removed the tissue from circulation and the immune system. It compared short-term bacterial survival under one ex vivo protocol, not healing, dosing, side effects or clinical outcomes.
The University of Nottingham’s initial announcement called the broader work an in vivo mouse model. The published methods give the essential detail for the headline comparison: the animals had been euthanized and the infected tissue excised before the four-hour soak.
No single ingredient explained the effect
Garlic looks like the obvious answer because crushing it produces allicin and other sulfur-containing compounds with antimicrobial activity. The experiments were more complicated. Individual ingredients did not match the complete recipe against established biofilms, and systematically removing ingredients changed the result.
Omitting garlic or wine eliminated bactericidal activity in the 2015 synthetic-wound assay. Removing onion or leek weakened it. Oxgall was necessary for the full activity of the onion version in one comparison, while removing brass made no measurable difference against S. aureus in that model.
Nine days mattered too. The authors proposed several possible explanations: wine may extract active molecules, ingredients may attack bacteria through different routes, or new compounds may form while the mixture stands. Their experiment showed that preparation altered activity, but it did not identify one definitive chemical mechanism.
A larger 2020 study in Scientific Reports reinforced the combination result. Across preparations made from 2014 to 2019, 62 of 75 onion-based batches cut viable S. aureus in biofilms by more than three orders of magnitude. Forty-five produced reductions greater than five orders.
The later work found activity against biofilms from several other wound pathogens, but not consistently against every species tested. The mixture was neither a one-ingredient story nor a universal antibacterial.
It remains a research lead, not a home remedy
Nothing in these results supports mixing the ingredients at home. The laboratory preparations controlled plant species, proportions, temperature, filtration and contamination. Homemade wine, bile and crushed vegetables are variable and nonsterile, and applying them to an infected eye could injure tissue or delay effective care.
Later safety work moved cautiously. A 2022 nonrandomized, open-label phase 1 study placed the preparation beneath a patch on healthy upper-arm skin in 109 volunteers for 48 hours. No serious adverse events were reported. That tested short-term tolerability on intact skin, not effectiveness against infection and not safety in the eye.
The broader idea is not unique to a medieval manuscript. ScienceBlog has also covered laboratory work on Manuka honey and MRSA. Natural mixtures can point researchers toward useful chemistry, but activity in a dish is the beginning of drug development rather than its conclusion.
Anyone with an eye or wound infection needs evidence-based assessment from a qualified clinician, not a homemade preparation.
The unresolved work is chemical and clinical: identify what the nine-day process produces, determine how the components interact, standardize the preparation, and test whether any resulting candidate can meet the safety and efficacy requirements of a medicine.
