Arnica, comfrey, plantain and St John's wort in traditional salves
Helenalin in arnica, pyrrolizidine alkaloids in comfrey, hypericin phototoxicity and plantain. Infusion ratios, use levels and the restrictions on each.
Arnica, comfrey, St John's wort and plantain turn up in almost every traditional salve book, usually with a ratio and no numbers behind it. Three of the four carry a constituent a regulator has an opinion about, and the fourth carries almost nothing into oil at all. This page gives what is measurably in each plant, how much of it reaches an infused oil, the restrictions that apply in the UK and EU, and the wording that keeps a finished product a cosmetic.
Arnica flowers must hold at least 0.40 percent sesquiterpene lactones to meet the European Pharmacopoeia, and the infused oil goes in at 5 to 25 percent, never on broken skin. Comfrey leaf carries roughly a tenth the pyrrolizidine alkaloids of root, so use leaf. St John's wort oil is red from hypericin. All four are cosmetic ingredients, not medicines.
What an oil infusion actually takes out of these plants
Oil is a non-polar solvent. It extracts what dissolves in fat: sesquiterpene lactones, triterpene acids, carotenoids, chlorophyll, cuticular waxes, resins and part of the plant's own volatile fraction. It leaves behind almost everything water soluble, which for this group of herbs means mucilage, polysaccharides, iridoid glycosides, tannins, saponins and most alkaloid salts. The general mechanism and the three working methods are in herbal infused oils.
That decides most of what follows. Arnica's active chemistry is lipophilic, so an oil infusion suits it and the risk travels with the benefit. Comfrey's headline compound is water soluble, so an infusion delivers little of it while still carrying alkaloids you would rather leave behind. Plantain and chickweed rest almost entirely on water soluble constituents, so an oil infusion of either is mostly green oil. Steeping longer changes none of this.
Arnica: helenalin, the monograph figure and the broken skin rule
The material is the dried flowerhead of Arnica montana, and the compounds that matter are pseudoguaianolide sesquiterpene lactones: helenalin, 11,13-dihydrohelenalin and their short chain esters. The European Pharmacopoeia monograph for Arnicae flos sets a minimum of 0.40 percent total sesquiterpene lactones on the dried drug, expressed as dihydrohelenalin tiglate. Commercial flowers commonly assay 0.3 to 1.2 percent, so a batch can fail the monograph and still be sold as a craft grade.
Chemotype matters more than most suppliers admit. Central European populations are dominated by helenalin esters, Spanish material by 11,13-dihydrohelenalin esters. Helenalin is the more cytotoxic and the more potent sensitiser, so two batches with identical total assays are not equivalent, and a certificate giving only a total does not tell you which you have.
Helenalin is also why arnica is one of the more frequently reported plant contact allergens. It is a Michael acceptor: the exocyclic methylene on the lactone ring reacts with protein thiols, the classical sesquiterpene lactone sensitisation mechanism. Sensitisation carries across the Asteraceae, which is why arnica, calendula, chamomile and yarrow sit together in the patch test sesquiterpene lactone mix.
Arnica preparations are for intact skin only. The traditional prohibition on broken skin, open wounds and mucous membranes is not folklore: helenalin is cytotoxic to keratinocytes and fibroblasts in culture, and damaged skin removes the barrier that keeps exposure low. A salve intended for use after a knock is also a salve someone will put on a graze. If you sell one, say "intact skin only" on the pack, not in a booklet.
Working level for the infused oil is 5 to 25 percent of the formula. The top of that range belongs in something used occasionally on a small area, such as the worked example in muscle rub balm; 10 to 15 percent is the better default for daily use. Pushing higher buys nothing, because the sensitisation risk scales with dose and the traditional use does not.
Comfrey: allantoin against pyrrolizidine alkaloids
Comfrey, Symphytum officinale and the hybrid Symphytum x uplandicum, is the awkward one. Its reputation rests on allantoin, carried in the root at roughly 0.6 to 4.7 percent of dry weight and in the leaf at a quarter to a half of that. Allantoin is also the thing an oil infusion cannot deliver: it dissolves at about 0.5 g per 100 mL in water at 25 C and is practically insoluble in vegetable oils, so a macerate holds an unknown but small quantity, well below any monograph level. If you want allantoin in the product, weigh in the powder at 0.2 to 1 percent, as set out in allantoin. Do not infuse comfrey and imply a dose.
What the oil does pick up, along with rosmarinic acid, triterpenes and chlorophyll, is a share of the plant's 1,2-unsaturated pyrrolizidine alkaloids: symphytine, echimidine, lycopsamine, intermedine and their N-oxides. These are hepatotoxic, and the concern is cumulative liver injury rather than acute harm. Concentrations differ by an order of magnitude between plant parts.
| Plant part | Allantoin | Total unsaturated PAs | Practical position |
|---|---|---|---|
| Root, dried | 0.6-4.7% | 200-2,000 mg/kg | Highest allantoin, highest alkaloids. The traditional choice and the risky one. |
| Leaf, dried mature | 0.3-1.5% | 30-200 mg/kg | Roughly a tenth the alkaloid load. The sensible default. |
| Leaf, young or spring growth | 0.3-1.5% | up to 500 mg/kg | Higher than mature leaf. Harvest later in the season. |
| PA-reduced or PA-free grade | varies | below stated limit | Selected cultivars or processed extract. Ask what the limit and method are. |
A supplier claim of "PA free" means nothing without a limit and a method. A credible one reads "below 0.35 mg/kg total 1,2-unsaturated pyrrolizidine alkaloids and N-oxides, sum of a named list, by LC-MS/MS". Anything vaguer is marketing, and your safety assessor will treat it as such.
Doing the pyrrolizidine alkaloid arithmetic
The numbers settle the leaf against root question immediately. Assume the worst case, that every alkaloid in the herb transfers into the oil, and follow it through to a daily applied dose.
| Starting material | PA in herb | PA in infused oil | PA in salve | Dose from 1 g/day, micrograms |
|---|---|---|---|---|
| Leaf, low end | 30 mg/kg | 6 mg/kg | 2.4 mg/kg | 2.4 |
| Leaf, typical | 100 mg/kg | 20 mg/kg | 8 mg/kg | 8 |
| Leaf, high end | 200 mg/kg | 40 mg/kg | 16 mg/kg | 16 |
| Root, typical | 1,000 mg/kg | 200 mg/kg | 80 mg/kg | 80 |
| Root, high end | 2,000 mg/kg | 400 mg/kg | 160 mg/kg | 160 |
Leaf at typical content leaves a margin of more than ten times. Root at the top of its range exceeds the reference figure outright, before you allow for anyone applying more than a gram or dosing several sites. That is the whole argument for leaf, and it rests on the ratio between the rows rather than on believing any particular threshold.
Two caveats, both making the table conservative. Complete transfer into oil is a deliberate overestimate, since PA N-oxides are usually the larger fraction in the plant and are too polar to extract well into a fatty carrier. And no robust dermal absorption fraction has been published for comfrey PAs, so the applied dose is not the absorbed dose. Redo it with your own ratio in the infusion calculator and a realistic figure from how much balm to apply.
What the rules actually say about PA botanicals
Most of the confusion here comes from quoting numbers that belong to medicines and food law rather than cosmetics law.
- Germany, medicines. The long standing reference limits for comfrey preparations are 100 micrograms of 1,2-unsaturated pyrrolizidine alkaloids and N-oxides per day externally, use restricted to four to six weeks a year, and 1 microgram per day internally. Almost every figure you will see quoted originates here.
- European Union, herbal medicines. The EMA's herbal committee set a transitional daily intake threshold for oral herbal medicinal products in the region of 1 microgram, since tightened. It applies to registered herbal medicines, not cosmetics.
- European Union, food. Maximum PA levels for herbal infusions, tea, borage leaves, cumin and supplements sit in the contaminants regulation, following the EFSA CONTAM panel's 2017 opinion. Food again, but it shows the direction of travel.
- Cosmetics. No entry in the annexes to Regulation (EC) No 1223/2009 or the UK Cosmetics Regulation sets a numeric pyrrolizidine alkaloid limit. Check the current CosIng entry before you rely on that, because the annexes are amended several times a year. Outside Europe, Health Canada's Cosmetic Ingredient Hotlist does restrict comfrey, so a product sold across borders may face a limit its home market does not impose.
The absence of a numeric cosmetic limit is not permission. In the UK and EU the control is the safety assessment: your assessor has to justify the exposure, and for a comfrey product they will want a measured PA figure on the raw material, your inclusion level and a realistic application amount. Turning up with "traditional use" and no number is how a safety assessment and CPSR stalls. The market obligations are in selling balms in the UK and EU.
St John's wort: hypericin, hyperforin and the red oil
The material is the flowering top of Hypericum perforatum, harvested at or just after full flowering. The European Pharmacopoeia monograph for Hyperici herba requires a minimum of 0.08 percent total hypericins, expressed as hypericin, on the dried drug. Two constituent groups matter and they behave completely differently in oil.
Hypericin and pseudohypericin are naphthodianthrones and they are what makes the oil red. They are only sparingly soluble in fatty oils, so the colour that develops over two to six weeks is partly hypericins and partly their photo-oxidation and degradation products. A deep red macerate is evidence that something transferred, not a measure of how much: published analyses of traditional oil macerates put hypericin content low, often in the single figures of mg per kg or below the limit of detection.
Hyperforin, a phloroglucinol present at 2 to 4.5 percent in fresh flowering tops, is the other headline compound. It is fat soluble and extracts readily, but it is also extremely unstable to light and oxygen, so most of it degrades during a multi-week maceration and the rest in storage. A home made St John's wort oil is a low hypericin, near zero hyperforin preparation with a strong colour.
The fresh against dried question is sharper here than for any other salve herb: the classic red develops best from fresh flowering tops, and fresh material is exactly the water hazard that makes an infused oil unsafe to store. The workable compromise is to wilt the tops in a single layer for twelve to twenty four hours, keep everything below the oil line, run the jar warm rather than sealed cold, then let the strained oil settle for two days and discard the bottom centimetre. If droplets rather than sediment appear, discard the batch. Or use properly dried tops, accept a paler oil, and lose nothing that has been shown to matter.
Phototoxicity: what the evidence supports
Hypericin is a genuine photosensitiser. The clearest evidence is veterinary: livestock grazing large quantities of Hypericum develop hypericism, a photodermatitis of unpigmented skin. In humans, severe phototoxicity has been documented at high intravenous and oral doses of purified hypericin in clinical trials, at exposures far above anything a herbal preparation delivers. At ordinary oral extract doses, reports of photosensitivity exist but are uncommon.
For a topical infused oil the honest position is that the risk is probably low and the evidence is thin. Hypericin concentration in a macerate is low, the salve goes on a small area, and no body of case reports links traditional St John's wort oil to phototoxic reactions. That is an absence of signal in a preparation nobody has studied properly, not a demonstration of safety.
What follows is a label instruction, not a ban. Tell users to avoid strong sun and sunbeds on treated skin, and flag it for people on photosensitising medication, people with light sensitive conditions, and anyone in the weeks after a laser or peel. The framework for that kind of warning is in essential oil allergy and photosensitivity, which covers the same reasoning for bergapten.
One thing you can drop from the warnings. St John's wort's famous drug interactions, the induction of CYP3A4 and P-glycoprotein that undermines oral contraceptives and immunosuppressants, are driven by hyperforin at systemic concentrations reached by swallowing standardised extract. A topical infused oil is not a meaningful systemic route and has almost no hyperforin left in it. Warning about interactions on a salve is inherited from the tablet and does not belong on the tin.
Plantain and chickweed: honest expectations
Plantain, Plantago major or P. lanceolata, is the herb most likely to disappoint in oil. Its studied constituents are the iridoid glycosides aucubin and catalpol, mucilage polysaccharides, tannins and flavonoids, all water soluble. The oil gets chlorophyll, cuticular waxes, a little ursolic and oleanolic acid and a modest amount of allantoin: a pleasant, mild green oil that is largely a colourant. Plantain's water soluble chemistry needs a water or ethanol preparation, and neither goes into an anhydrous balm.
Chickweed, Stellaria media, is worse. Fresh it is over ninety percent water, so a kilogram dries to under a hundred grams, and its saponins and vitamin C are water soluble and heat labile. The oil takes up chlorophyll and very little else. Not harmful, not much of an ingredient.
Neither is a reason to drop them from a traditional formula. Both are safe, cheap and green. Just recognise that a blended salve carrying them is doing its work through the occlusive fat and the wax, not the botany, which is the point made throughout herbal salve.
Ratios, drying and straining, herb by herb
| Herb | Ratio by weight | Method and handling | Oil in the salve |
|---|---|---|---|
| Arnica flowerheads | 1:8 to 1:10 | Very low bulk density, so a 1:5 will not stay submerged. Filter twice: the pappus hairs pass a coarse muslin and irritate. | 5-25% |
| Comfrey leaf | 1:5 to 1:6 | Crumble dry leaf before covering. Hairy leaf traps air, so stir out the bubbles on day one. | 20-50% |
| Comfrey root | 1:5 | Chop or grind before drying, it goes rock hard. Needs the 50 to 70 C method, cold maceration barely touches it. | under 20% |
| St John's wort tops | 1:5 dried, 1:3 wilted | Wilted material carries residual water, hence the tighter ratio and the mandatory settle and decant. | 10-30% |
| Plantain leaf | 1:5 | Dries fast and cleanly. Nothing difficult about it. | up to 100% of the oil phase |
| Chickweed | 1:5 | Dry thoroughly, it holds water longer than it looks. Expect a large mass loss. | up to 100% of the oil phase |
Drying standard is the same for all six: brittle, not leathery. Leaves crumble between the fingers, stems snap, roots break rather than bend. A herb that is nearly dry is the one that spoils a batch three weeks later. Choose an oxidatively stable carrier too, since these infusions run for weeks: the ranking is in carrier oils compared.
Strain through two layers of muslin, press, then leave the oil undisturbed in a tall jar for 24 to 48 hours and decant off the top, discarding the last centimetre. Fines and trace water settle out, and that one step is the difference between an oil that keeps a year and one that turns in three months. Label the jar with the herb, the plant part, the ratio, the carrier and the date, because every infused oil looks identical by February.
Where the cosmetic claim stops and the medicine begins
None of these four herbs is banned from cosmetics in the UK or EU. The claim is what gets a maker into trouble, because under the Human Medicines Regulations 2012 a product becomes a medicine by function or by presentation, and presentation means how you describe it. MHRA guidance is explicit that an ingredient with a pharmacological reputation plus a claim aimed at a condition produces a medicine whatever the label calls itself.
These herbs sit on that line, and the proof is that topical arnica and comfrey products hold Traditional Herbal Registrations in the UK. The regulator has already decided those claims are medicinal. Repeating them on a cosmetic is not a grey area, it is the same claim without the registration.
So: bruising, sprains, strains, inflammation, healing, wound care, muscle injury and joint pain are medicinal claims. Softening, conditioning, comforting, protecting, keeping skin in good condition and perfuming are cosmetic ones. "Traditionally used by herbalists for" does not launder a medicinal claim, and neither does putting it on your website rather than the pack. Work through cosmetic versus drug claims before you write any copy, and read drawing salves and black salve for what happens at the far end of this spectrum.
What to print, and where each herb is the wrong choice
Label wording that stays cosmetic and still tells the truth: intact skin only; not for use on broken skin, open wounds or mucous membranes; if you are allergic to daisies, marigolds or ragweed, patch test before use; avoid strong sunlight and sunbeds on treated skin (St John's wort); not recommended during pregnancy or breastfeeding (comfrey); discontinue if irritation occurs. Patch testing advice for customers is in how to patch test a balm, and the mandatory pack elements are in labelling cosmetics in the UK. Anything beyond that belongs in the general limits set out in the safety disclaimer.
The decision rule, herb by herb. Drop arnica if the product is aimed at hands, feet or anywhere a customer will use it daily on skin that is sometimes cracked, and drop it entirely from anything for children. Use comfrey leaf, never root, and if you cannot get a measured PA figure on the raw material, do not make the product at all: the worked version is in comfrey salve. Use St John's wort only where the deep red colour is part of the product, since that is the one thing you can be confident you extracted. Treat plantain and chickweed as green, mild and traditional rather than active.
Keep the category in proportion. A herbal salve works mainly because it is an anhydrous occlusive layer that slows water loss, which is a property of the fat and the wax rather than the herb. The infusion adds colour, scent, a little lipophilic plant chemistry and a story. If you want measurable performance, the leverage sits in the base, in a dosed active, or in a warming agent of the kind covered in menthol, camphor and actives. The herbs are the part easiest to get wrong and hardest to prove right.
Frequently asked questions
Should I use comfrey leaf or root in a salve?
Leaf. Root carries roughly ten times the pyrrolizidine alkaloid load, at 200 to 2,000 mg per kg against 30 to 200 for mature leaf. On a worst case calculation, a salve made with root oil can approach or exceed the 100 microgram per day external reference limit used for comfrey preparations, while a leaf version sits ten times below it. Harvest mature leaf, not spring growth.
Why should arnica never go on broken skin?
Its sesquiterpene lactones, principally helenalin, are cytotoxic to skin cells in culture and are potent contact sensitisers because the lactone ring reacts with protein thiols. Intact skin limits exposure; a graze or crack does not. The instruction belongs on the pack, not in a leaflet, because a salve sold for use after a knock will be used on grazes.
Is St John's wort oil phototoxic on the skin?
Hypericin is a real photosensitiser, but the evidence for topical infused oil causing reactions in humans is thin rather than reassuring. Concentrations in a macerate are low, often single figures of mg per kg. Print a warning to avoid strong sun and sunbeds on treated skin, and flag it for people on photosensitising medication.
Does comfrey salve deliver allantoin?
Barely. Allantoin dissolves at about 0.5 g per 100 mL in water and is practically insoluble in vegetable oils, so an oil infusion extracts very little of it whatever the ratio or steeping time. If you want a known amount of allantoin in a balm, weigh in the purified powder at 0.2 to 1 percent and disperse it as a suspension.
Are these herbs legal in cosmetics in the UK and EU?
The plants themselves are not prohibited, and no numeric pyrrolizidine alkaloid limit appears in the cosmetics annexes. The control is the safety assessment, which requires you to justify the exposure with real figures, and the claim, since describing a product as treating bruises, sprains or inflammation makes it a medicine by presentation regardless of the label.
How much infused herb oil should go into a salve?
Arnica oil at 5 to 25 percent of the finished formula, comfrey leaf oil at 20 to 50 percent, St John's wort oil at 10 to 30 percent, and plantain or chickweed oil up to the whole of the liquid oil phase. The rest of the formula is wax and butter, which is where the actual occlusive performance comes from.
Sources and further reading
- European Directorate for the Quality of Medicines, European Pharmacopoeia monographs: Arnicae flos (arnica flower) and Hyperici herba (St John's wort), Strasbourg.
- European Medicines Agency, Committee on Herbal Medicinal Products, Public statement on the use of herbal medicinal products containing toxic, unsaturated pyrrolizidine alkaloids, Amsterdam.
- European Medicines Agency, Committee on Herbal Medicinal Products, Assessment report on Symphytum officinale L., radix, Amsterdam.
- European Food Safety Authority, CONTAM Panel, Risks for human health related to the presence of pyrrolizidine alkaloids in honey, tea, herbal infusions and food supplements, EFSA Journal, Parma.
- Commission Regulation (EU) 2023/915 on maximum levels for certain contaminants in food, EUR-Lex. Pyrrolizidine alkaloid limits for herbal infusions and supplements.
- Medicines and Healthcare products Regulatory Agency, Guidance Note 8: a guide to what is a medicinal product, London.
- European Commission, CosIng cosmetic ingredient database, entries for Arnica Montana Flower Extract, Symphytum Officinale Leaf Extract and Hypericum Perforatum Extract.
Reviewed and updated 6 September 2026. Spotted an error? Tell us and we will fix and log it.