Shelf life and stability testing for balms: what to measure and how to date a batch

What really limits a balm's life, what induction periods and peroxide values do and do not tell you, and a stability protocol for a small maker.

A balm does not spoil in the way food spoils. It has no water in it, so nothing grows, and the failure it does suffer is chemical and gradual. That changes what stability testing means: you are not looking for a microbial event, you are watching a slow reaction and deciding when its products become unacceptable. This page covers what is actually measurable, what is vendor guidance, and how to turn both into a date you can defend.

What actually limits a balm's life

Two failure modes compete, and only one of them matters in an anhydrous product. Microbial spoilage needs available water. A genuine balm does not supply it, which is why balms do not need preservatives. Oxidative failure needs only oxygen, and it proceeds happily in a dry system. Every unsaturated fatty acid in the tin is slowly reacting with atmospheric oxygen, forming hydroperoxides that then break down into the aldehydes you can smell. The full mechanism is set out under rancidity and oxidation; here we are concerned with predicting and dating it.

There is a third failure mode worth naming because it fails a batch just as effectively without any chemistry at all: physical instability. Grain, sweating, separation, shrinkage and colour drift will make a product unsellable long before its peroxide value becomes interesting. A stability protocol has to look for all three.

The bis-allylic count predicts rancidity, not the double-bond count

Autoxidation starts with hydrogen abstraction, and that step is rate limiting. The vulnerable hydrogens sit on a bis-allylic methylene, a CH2 flanked by two double bonds. Losing one gives a pentadienyl radical delocalised over five carbons, far more stabilised than the allyl radical you get from a single double bond. The bond dissociation energies differ by roughly 10 kcal/mol, near 77 for bis-allylic against near 87 for allylic.

This is the single most useful predictive idea available to a formulator. Oleic acid is unsaturated but has no bis-allylic hydrogen, so on a balm's timescale it behaves almost like a saturate. Jojoba is a mixture of monounsaturated wax esters with no bis-allylic position anywhere and no glycerol backbone to hydrolyse, which is why it outlives every triglyceride oil on the bench. A workable index for a blend is percent linoleic plus twice percent linolenic, weighted by each oil's share. Rank your formulas by that number before you test anything.

Relative rates depend on what you measure, so quote a range rather than a single figure. Oleic to linoleic to linolenic comes out at 1 : 40-50 : 100 by oxygen uptake, 1 : 12 : 25 by hydroperoxide formation, and 1 : 70 : 265 by absolute propagation rate constants at 303 K. The widely circulated "1 : 100 : 1200 : 2500" cannot be traced to a primary source and is an order of magnitude away from all of these. It should not be repeated.

Measured induction periods, and why they are not months

Induction period is the time before oxidation accelerates sharply, measured under forced conditions. Four cosmetic oils tested under identical conditions at 100 C gave a clean ranking that tracks polyunsaturate content almost perfectly.

Jojoba lasts about nine times longer than rosehip under identical stress. Other measured values in the same literature put jojoba at 68.6 to 71.2 hours at 110 C, and at 51 to 229 days at 60 C against 1 to 29 days for every other oil tested, with meadowfoam at 15 hours at 130 C.

Packaging and temperature beat oil choice

This is the headline finding, and it is the one most maker guidance gets backwards. A 270-day real-time study followed the same hemp seed oil, one of the least stable oils in common use, under different storage conditions. In polypropylene at 25 C in diffused light it reached 40 meq/kg by day 150, nearly three times the Codex limit for a cold pressed oil. In amber glass with a foil overwrap at 10 C in the dark it stayed under 10 meq/kg for the full 270 days. Same oil, same starting material, opposite outcomes.

Read that carefully. The variable that failed the oil was not its fatty acid profile, which was identical in both arms. It was light, temperature and container. A maker who swaps hemp for high-oleic sunflower but stores stock in clear plastic on a warm shelf has optimised the weaker lever and ignored the stronger one. Rank the levers in this order.

1. Bis-allylic content of the blend. Chosen once, at formulation.
2. Storage temperature and light exposure. Cool, dark, and never a sunny window or a warm car.
3. Headspace oxygen and packaging. Full containers, opaque or amber where possible, tight closures. See the packaging guide for material choices.
4. Native tocopherol content. Real but modest, and covered under vitamin E and antioxidants.
5. Trace metal contamination. Controlled by buying well and avoiding prolonged unlined metal contact, not by adding a chelator to a water-free base.

Peroxide value, read properly

Peroxide value is measured by iodometric titration and expressed in milliequivalents of peroxide oxygen per kilogram of oil, under ISO 3960 or AOCS Cd 8b-90. It is the number you should ask a supplier for, and it is routinely misread.

The gap between those last two rows is the practical point. A certificate showing 8 meq/kg is legal and is not fresh. You are buying oil that has already spent a meaningful part of its life somewhere warm, and your finished balm inherits that history. Ask for the peroxide value and the date of analysis at the point of purchase, which is part of what sourcing ingredients properly means.

The second misreading is more serious. Peroxide value is not monotonic. Hydroperoxides are intermediates: they form, then they decompose into the aldehydes and ketones you can smell. A badly rancid oil can therefore return a low or falling peroxide value, because its peroxides have already broken down. A single reading is uninterpretable. It means something only when plotted over time and paired with a secondary measure such as p-anisidine value. Acid value, by contrast, rises throughout, which makes it a useful companion. Totox, defined as twice the peroxide value plus the anisidine value, is often quoted against thresholds of under 10 as good, 10 to 20 moderate and over 20 advanced, but those bands are trade convention rather than a standard, and should be described as such.

A stability protocol a small maker can actually run

You do not need a laboratory. You need retained samples, a cheap oven or incubator, a freezer and a diary. Run the protocol on every new formula and on any formula where you change an oil, a supplier or a container.

The monthly check is the part people skip and the part that matters. Take the ambient retained sample and the frozen control, let both reach room temperature, and compare them directly: smell each on the back of the hand after ten seconds, look at colour against white paper, press the surface for hardness and note any sweat, grain or dip. Write down what you observe even when nothing has changed, because an unbroken run of "no change" entries is exactly the evidence that supports a date.

Treat the 40 C oven as a failure-finding and comparison tool, not a date generator. It reliably tells you that formula A separates and formula B does not. It cannot tell you that twelve weeks at 40 C equals two years at ambient, for the same reason induction periods cannot be converted: the extrapolation is not valid for lipid oxidation.

Setting an honest best-before date

Work from the constraint, not from the hope. The shortest-lived oil in the blend sets the ceiling, and vendor shelf-life figures for oils are estimates with no stated method, temperature, packaging or endpoint, so treat them as guidance rather than data. The comparison table under carrier oils compared gives the ranges Trebalm uses.

1. Take the shortest-lived component. A blend containing 10 percent rosehip is a six-month blend, not a two-year one, however stable the other 90 percent is.
2. Subtract the age of your stock. Oil that was six months old when you made the batch has already spent that time. Date from pressing where you can, not from delivery.
3. Apply a margin. Claim less than your ceiling. Twelve months on a blend you believe will hold eighteen is a sensible position for a product you cannot control after it leaves you.
4. Confirm with retained samples. Never claim a date that runs past the point your own retained samples have reached and passed.
5. Record the reasoning. One page per formula: oils, ages, protocol results, date claimed. That page is your evidence if anyone asks.

Labels, PAO and the 30-month rule

The date on the pack is a legal statement, not a marketing one. In the EU and UK, Regulation (EC) No 1223/2009 Article 19 requires a date of minimum durability, expressed as best before end, on any product whose minimum durability is 30 months or less. Products with a minimum durability of more than 30 months are exempt from that date and must instead carry the period-after-opening symbol, the open jar marked 6M, 12M or similar, where the concept is meaningful for the product. Most handmade balms fall under the 30-month line and therefore need a real date. The wider requirements are covered under selling balms in the UK and EU.

US practice differs: there is no general federal requirement to put an expiration date on a cosmetic, but the product still has to be safe throughout its intended life, and the responsibility for that sits with the manufacturer. Voluntary dating is common and sensible. See labelling cosmetics in the US for what must appear.

The microbiological side, briefly but correctly

ISO 29621 lists physical and chemical factors that can make a cosmetic product microbiologically low risk. Water activity below 0.75 is one of them, and a high filling temperature above 65 C is another, with lip balms and lipsticks named as the example. Meeting one of those criteria means challenge testing to ISO 11930 is generally not necessary.

It does not mean testing is finished. Enumeration testing to ISO 17516 is still expected for a product sold commercially, because water activity does not correlate with microbial load: a raw material can arrive carrying a high count of dormant organisms and spores, and the standard's own position is that any exemption must rest on appropriate justification supported by reliable data. Note also that hot filling is the weakest criterion in the list, because it reduces the count at the moment of filling and does nothing about recontamination afterwards. The controls that actually keep counts low are upstream, in the workshop, and are covered under workshop hygiene.

What to do next

Pick one formula you already sell and start its retained samples this week: one frozen, one ambient in final packaging, one in a 40 C oven, all three dated and logged. In three months you will know more about that product than any supplier table can tell you, and in twelve you will have an evidence-backed date instead of an inherited one. Then work backwards through your stock and ask every supplier for a peroxide value with a date of analysis attached. Those two habits, retained samples and incoming testing, do more for shelf life than any antioxidant you can add.

Reviewed and updated 6 September 2026. Spotted an error? Tell us and we will fix and log it.