A large share of the oils, esters and fatty acids sold into cosmetics, food and lubricants arrive with three oxidation numbers on the certificate: peroxide value (PV), p-anisidine value (p-AV) and, on the better certificates, TOTOX. Buyers tend to treat them as three measurements of the same property and to specify whichever one appears on the template. They are not the same property. They sit at three different points on one reaction chain, they are produced by three methods whose reproducibility differs by an order of magnitude, and the arithmetic that combines them has a failure mode that neither individual limit can catch.
One reaction chain, three windows onto it
Lipid oxidation starts with an unsaturated fatty acid losing a hydrogen next to a double bond. The resulting radical takes up oxygen and becomes a hydroperoxide (LOOH). Hydroperoxides are the primary products, and they are unstable: with heat, light or trace metals they cleave into aldehydes, ketones, alcohols and short-chain acids — the compounds that actually smell rancid. Oxidation therefore has a shape, and each index samples a different part of it:
- Fresh material: low PV, low p-AV.
- Early oxidation: PV climbing, p-AV still near zero. Nothing smells wrong yet.
- Advanced oxidation: PV may already be falling because the peroxides have decomposed, while p-AV climbs steeply.
- Rancid: PV low, p-AV high. The oil is wrong in the nose and near-perfect on a PV-only specification.
That last row is the one that costs money. An oil reading PV 3 meq/kg with p-AV 35 is not fresh; it is a sample whose primary products have already broken down. PV is a peak-shaped curve, not a progress bar. This is also why the method itself describes PV as a dynamic parameter whose value depends on the history of the sample — the number tells you where the oil is now, not where it has been.
The unit trap comes first
Before any comparison, confirm which unit the certificate is using. Peroxide value is conventionally reported in milliequivalents of active oxygen per kilogram, but ISO 3960 explicitly allows the SI expression in millimoles, and the two differ by a factor of two:
| Expression | A PV of 10 meq/kg reads as | Conversion |
|---|---|---|
| meq active oxygen / kg (conventional) | 10 | reference unit |
| mmol active oxygen / kg (SI) | 5 | divide meq by 2 |
| mg active oxygen / kg | 80 | multiply meq by the equivalent mass of oxygen, 8 |
A specification that says "peroxide value, max 10" without a unit is therefore ambiguous by a factor of two, and a purchase order that crosses a lab using meq/kg with one using mmol/kg will reject good lots and pass bad ones. This is the cheapest error in the whole document to prevent and the most common one to find.
Peroxide value: iodometric, empirical, and mass-dependent
The reference methods are ISO 3960 (iodometric, visual endpoint) and AOCS Cd 8b-90; a potentiometric alternative exists as ISO 27107. The sample is dissolved in isooctane and glacial acetic acid, potassium iodide is added, and the liberated iodine is titrated with sodium thiosulfate. ISO 3960 covers readings from 0 to 30 meq/kg, which is comfortably above any purchasing limit.
Two properties of the method matter commercially. First, it is a highly empirical procedure: the standard warns that the value obtained depends on the sample mass, and that the prescribed mass yields slightly lower values than a smaller one — which is why the method requires the sample mass to be recorded with the result. Second, its precision is not what a spec sheet implies. Repeatability is a few percent, while inter-laboratory reproducibility sits in the region of 14% relative standard deviation. On a limit of 10 meq/kg that is roughly 1.4 meq/kg of pure laboratory noise, so a supplier reporting 11.2 and a customer measuring 9.4 are describing the same drum.
The regulatory anchor for edible oils is Codex CXS 210-1999: a maximum of 10 meq of active oxygen per kg for refined oils and 15 meq/kg for cold-pressed and virgin oils, alongside acid value limits of 0.6, 4.0 and (for virgin palm oil) 10.0 mg KOH/g.
p-Anisidine value: the number that never goes back down
The p-anisidine value is defined by convention as 100 times the optical density at 350 nm of a solution of 1 g of oil in 100 mL, measured after reaction with p-anisidine in acetic acid (AOCS Cd 18-90, ISO 6885). Operationally, p-AV = 25 × (1.2 × As − Ab) / m, where As is the absorbance after reaction and Ab the absorbance of the oil solution alone.
It detects aldehydes, principally 2-alkenals and 2,4-dienals. A double bond conjugated with the carbonyl raises the molar absorbance four to five times, which makes the test very sensitive to exactly the aldehydes that come from oxidised linoleic and linolenic chains — and insensitive to saturated aldehydes that may matter just as much to the nose.
Three limits of the method should be written into any decision made from it:
- It is not specific. p-Anisidine reacts with any aldehyde, so a flavoured oil (citral, vanillin, citrus terpenes) can read high with no oxidation at all.
- It fails on coloured oils. Carotenoids and other pigments absorb near 350 nm; the method cannot be used without a blank correction, and for some matrices not at all.
- It should not be used above PV 5. p-Anisidine also reacts slowly with hydroperoxides, so it does not cleanly separate primary from secondary products. Several method reviews advise against applying it to oils with PV above 5 meq/kg — a caution that is widely ignored.
And its precision is the worst of the three: published collaborative data for AOCS Cd 18-90 on samples around p-AV 2 give repeatability coefficients of variation of 4–6% but inter-laboratory reproducibility of 30–37%. Two labs can disagree by a third on a low p-AV and both be correct.
TOTOX: the arithmetic that catches a lot passing both specs
TOTOX = 2 × PV + p-AV. The peroxide term is doubled because hydroperoxides are active precursors that will decay into further secondary products, so they are weighted as future damage rather than current damage.
The reason it exists is a pure arithmetic gap between two separate limits. Take a specification of PV ≤ 5 meq/kg and p-AV ≤ 20, the GOED voluntary limits for marine oils, which Codex adopted for fish oils. A lot with PV 5 and p-AV 20 passes both columns. Its TOTOX is 30 — above the 26 limit, and by construction a lot sitting at both maxima at once is far more oxidised than either single limit intends. This point was put explicitly on the record during the Codex discussion of fish oil standards: TOTOX must be set separately and lower than the sum of the individual limits, otherwise the two limits can both be satisfied by an unacceptable oil.
Useful anchor points: PV ≤ 5, p-AV ≤ 20, TOTOX ≤ 26 for marine oils; and for a genuinely fresh oil, p-AV below 2 and TOTOX below 4. Note also that the 26 limit belongs to a product family — it is a Codex/GOED marine oil figure, not a universal threshold for a cosmetic ester.
Rancimat / OSI: an accelerated test, not a shelf life
ISO 6886 (identical in scope to AOCS Cd 12b-92) heats the sample to a fixed temperature, usually 100–120 °C, passes air through it at a controlled flow, and carries the volatile acids into a conductivity cell. The time until conductivity rises sharply is the induction period, reported in hours as the oxidative stability index.
The working rule is that a 10 °C rise roughly halves the induction period — ISO 6886 states the factor as approximately 2 — and the optimal induction period is 6–24 h. Extrapolation is therefore arithmetic: an OSI of 10 h at 110 °C corresponds, on a factor of 2 per 10 °C, to 28.5 ≈ 360 times that at 25 °C, or roughly 3,600 h, about five months. That figure is an order-of-magnitude sanity check, nothing more. The standard is explicit that the method does not allow determination of stability at ambient temperature; it exists to rank samples and to compare antioxidant efficacy under identical conditions.
Its precision is coarse at the inter-laboratory level: repeatability within 6% of the mean, reproducibility within 29%. Two labs reporting 8.0 h and 10.4 h on the same oil are reporting the same oil. The legitimate use is comparative — 0.02% of one antioxidant against 0.02% of another, same instrument, same day, same temperature.
Acid value is hydrolysis, not oxidation
Acid value is frequently read as an oxidation number and is not one. It measures free fatty acids released by hydrolysis — water plus heat, or a lipase — not by oxygen. The two failures smell different, look different and are prevented differently: oxidation is fought with antioxidants, chelators and oxygen-free packing; hydrolysis is fought with dry raw material, low temperature and short dwell time.
The conversion is worth memorising: FFA% (as oleic) = acid value × 0.503, from the ratio of oleic acid (282.46) to KOH (56.11) over 10. An acid value of 2.0 mg KOH/g is 1.0% free oleic acid. For an ester, acid value rising over time in storage is the signature of hydrolysis; for an oil, it is the signature of a wet or enzymatically active feedstock.
Which number governs which substrate
| Substrate | Governing index | Why |
|---|---|---|
| Caprylic/capric triglyceride (CAS 65381-09-1) | PV, then p-AV on aged stock | Fully saturated C8–C10 chains: no allylic hydrogen, so oxidation is slow and OSI is long. Odour complaints usually come from the C8 fraction, not from oxidation. |
| Oleic acid (CAS 112-80-1) | PV, p-AV and acid value together | One double bond: oxidation is slow but real; colour and odour drift track p-AV, not PV. |
| Castor oil (CAS 8001-79-4) | Acid value first | Ricinoleic acid carries a hydroxyl group; the practical failure in storage is hydrolysis and viscosity drift, not rancidity. |
| White mineral oil (CAS 8042-47-5) | None of the three | Saturated hydrocarbon, essentially no oxidation pathway. The governing numbers are UV absorbance, carbon type and aniline point. |
| Tocopherol (CAS 10191-41-0) | Assay, plus the PV of the oil it protects | It is consumed while working; the useful question is how much is left after processing, not how much was added. |
| Tocopheryl acetate (CAS 7695-91-2) | Assay only | The acetate group blocks the phenolic hydroxyl, so it cannot donate a hydrogen to a lipid radical. It is a stable vitamin E claim, not an oil protectant. |
| BHT (CAS 128-37-0) | Residual assay by GC or HPLC | Volatile and consumed; the compliance limit (EU E321: 100 mg/kg in fats and oils) is about what is present, and the stability question is how much survives. |
Dosing notes that the indices make obvious
- More tocopherol is not better. Measured optima in bulk oil sit around 100 mg/kg for alpha-tocopherol and around 300 mg/kg for gamma; above the optimum the tocopheroxyl radical it forms can initiate new chains, and the oil oxidises faster with more antioxidant in it. Add it to the oil phase below 50–60 °C, as late as the process allows.
- Tocopheryl acetate protects nothing. Direct radical scavenging in a bulk lipid is essentially zero before enzymatic hydrolysis; substituting the acetate for the free tocopherol in an oil-protection role is a straightforward procurement error, and a costly one at the acetate's price.
- Chelators are part of the oxidation specification. Trace iron and copper catalyse hydroperoxide decomposition at parts-per-million levels; citric acid (CAS 77-92-9) at a few hundred mg/kg is the usual answer, and it works on the metal, not on the radical.
- Antioxidants retard; they do not reverse. Both 21 CFR 170.3 and EU Regulation 1333/2008 Annex I define them as retarding or prolonging. Once p-AV is up, no addition brings it down — only dilution does, and dilution is a labelling decision.
What to write on the purchase order
- Method and unit on every limit: "PV ≤ 10 meq active oxygen/kg, ISO 3960", not "PV max 10".
- Sample mass reported with the PV, because the method says the result depends on it.
- p-AV only where it is valid: specify it for pale, unflavoured oils and esters; skip it for coloured or flavoured matrices, where it measures the flavour.
- TOTOX as a third, independent limit for marine and highly unsaturated oils, set below the sum of the two individual limits.
- OSI with temperature and instrument: "OSI ≥ X h at 110 °C, ISO 6886" — and compare only against the same substrate and the same temperature.
- Packaging and headspace: nitrogen blanket, tight headspace, opaque or amber containers; light and oxygen are the two variables the buyer still controls after the drum is filled.
Supply
For the substrates discussed above we supply caprylic/capric triglyceride (GTCC) (CAS 65381-09-1), caprylic/capric/succinic triglyceride (CAS 97708-73-1), oleic acid (CAS 112-80-1), erucic acid (CAS 112-86-7) and castor oil (CAS 8001-79-4); the antioxidant package side with tocopherol (CAS 10191-41-0), dl-alpha-tocopheryl acetate (CAS 7695-91-2), BHT (CAS 128-37-0) and BHT food grade (CAS 128-37-0, same substance, different compliance pack), plus citric acid anhydrous (CAS 77-92-9) as the chelator. For the saturated-hydrocarbon baseline we stock white mineral oil 32# (CAS 8042-47-5).
For the vitamin E forms and the IU arithmetic behind them, see our tocopheryl acetate IU conversion guide; for how antioxidant chemistry maps onto failure mechanisms, see our antioxidant selection by failure mode; and for the purity grades of the mineral oil baseline, our white mineral oil refining depth guide. All grades are available for B2B supply with sample quantities shipped worldwide — requests for COA interpretation, oxidation specifications or stability data are welcome through the enquiry form on any product page.