Cloudy balm that spat in the pot, and how to find where the water came from
Water droplets scatter light, spit on heating and create pockets where microbes grow. Drying temperatures, condensation control and when to bin the batch.
An anhydrous product is defined by the absence of one ingredient, so the whole safety argument collapses the moment water gets in. A balm that hazes over, crackles in the pot or sets milky is telling you it happened. This page is about reading those signals correctly, tracing the water to its actual source, and deciding whether the batch in front of you can be dried or has to go in the bin.
Haze, spitting above 100 C and a milky set all mean free water. A dispersed droplet is water at an activity of 1.0 no matter what the bulk reads, and growth is possible from about 0.6 upwards. Dry infused oils and rendered fats at 100 to 110 C until the crackling stops. Only redry a batch that is under a week old, unopened and unspoiled.
Three signals, and what each one measures
Water announces itself in three ways, and they are not interchangeable. Each one is sensitive to a different quantity of water, which is why a batch can pass one check and fail another.
| What you see or hear | The physics | Water it implies | What to do next |
|---|---|---|---|
| Crackling or spitting as the pot passes 100 C | Dispersed droplets flashing to steam and blowing through the oil above them | Roughly 0.05% and up | Keep heating until it goes silent, then hold |
| Milky haze in the hot melt that will not clear | Fine droplets scattering visible light | Roughly 0.1% and up | Dry the batch, do not just remelt it |
| Dull, opaque or chalky set where you expected translucency | Droplets locked in place by the crystallising fat network | Roughly 0.1% and up | Confirm with the heat test before blaming crystals |
| Clear beads or a layer under the balm in a jar | Water that has coalesced and settled, being the denser phase | Above about 0.5% | Decant, then dry what is left |
| Droplets on the underside of a lid | Condensate from capping a warm product, or from thermal cycling | Variable, and it keeps arriving | Fix the cooling and storage routine |
| Fuzzy, raised or coloured growth on the surface | Not a water symptom any more, but its consequence | Enough, for long enough | Discard, and read mould in balm |
The last row is the reason this page exists. Everything above it is cosmetic damage that can usually be reversed. The moment water and time produce growth, the batch is finished, and no amount of heat makes it saleable again.
The spitting test is a laboratory method in miniature
The free diagnostic anyone can run is also, almost exactly, the official one. The hot plate method for moisture and volatile matter in fats and oils (AOCS Official Method Ca 2c) works by heating a weighed sample until the bubbling and crackling stop, then weighing what is left. The crackle is the endpoint indicator, and the water content is the mass difference. Your ear is doing the same job as the balance, just without the number.
The noise itself is worth understanding. Water boiling under a layer of oil does not simmer politely. At 100 C and atmospheric pressure, liquid water expands to about 1,600 times its volume as steam, so a droplet a fraction of a millimetre across becomes a bubble that throws hot oil out of the pot. That is why the test is loud, and why it is also the main hazard on this page.
Molten fat at 110 C causes deep scalds and clings to skin. Test in a wide, shallow pan filled no more than a third, use a splatter guard rather than a lid, keep your face away from the surface, and never test a large batch to find out whether a small one is wet. A lid is the wrong tool because it condenses the steam and returns it to the oil, which is the exact failure mode that produces wet infused oils in the first place.
Run it on 20 to 30 g in a small stainless pan, raise the temperature steadily with an immersed probe, and listen from about 95 C. Silence through 110 C means no significant free water. Crackling means water, and the volume of it tracks roughly with the amount. The honest limit is sensitivity: this finds free water at around 0.05 percent and upwards, while a Karl Fischer titration (ISO 8534) resolves parts per million. Silence is not proof of dryness in the analytical sense, only proof that there is no pool of water for an organism to live in.
Why a wet balm looks white
Cloudiness is not a chemical change, it is optics. A droplet of water suspended in fat has a different refractive index from the fat around it, so light bends at every interface it crosses. When the droplets are similar in size to the wavelength of visible light, roughly 0.4 to 0.7 micrometres, scattering is strong and nearly the same at every wavelength, which is why the result is white rather than coloured. Milk is white for exactly this reason, and so is a wet balm.
Droplet size explains the rest of the behaviour. Above about 5 micrometres, droplets coalesce and sink, which is why gross contamination shows as a puddle rather than a haze. Below about 0.05 micrometres, they scatter too little to see, so a product can be visibly clear and still carry water. Between the two, the balm is milky.
The set is where it becomes permanent. In a liquid oil, droplets drift and merge, so a wet oil left standing for a day or two often separates and can be decanted. In a balm, the wax and butter network crystallises around the droplets and immobilises them, exactly as it immobilises the liquid oil phase described in oleogels and wax networks. Nothing can move, nothing coalesces, and the haze is locked in. Remelting to 70 C and repouring will not clear it, because 70 C removes no water. Only taking the melt past 100 C does.
Where the water came from, in order of likelihood
Trace it by size of contribution rather than by ease of blame. The first three sources on this list account for the great majority of cases and all of them are ingredient problems rather than accidents at the bench.
| Source | Typical water | How it gets into the balm |
|---|---|---|
| Fresh or wilted botanicals | 70-90% of the plant | Infused directly, the sap leaves the tissue and stays in the oil |
| Air dried herbs that are not fully dry | 8-12%, leathery ones more | Slow release into the oil during a long infusion |
| Infused oil dried in a partly covered pan | Variable, often 0.5-2% | Steam condenses on the lid and runs straight back in |
| Wet rendered tallow or lard | Up to several percent | Water used in rendering stays emulsified in the fat |
| Condensation on cold containers or a cold lid | Small but repeated | Deposited every time the product crosses its dew point |
| Damp utensils and rinsed, undried jugs | A few tenths of a gram | Carried in on the tool, then dispersed by stirring |
| Deliberate additions | Honey about 17-20%, aloe and hydrosols above 95% | Formulated in, which makes the product one that needs preserving |
| Refined oils and butters as supplied | 300-1000 ppm | Dissolved rather than dispersed, and not usually the problem |
The last row deserves a note, because it is where most anxiety lands and it is the least important line in the table. Vegetable oils carry a few hundred parts per million of water as bought, dissolved in the lipid rather than sitting as droplets. That is normal, it is not what makes a balm cloudy, and it is not what feeds a mould. The problem is always dispersed liquid water, and the routes that supply it are the seven rows above. Deliberate additions are a different conversation entirely, because honey in balms or a hydrosol makes the product a preserved one by definition.
The routing question also decides your corrective action. If the water arrived with an ingredient, the fix is upstream in herbal infused oils or in your rendering. If it arrived after filling, the fix is in workshop hygiene and packaging. Those are different projects and confusing them wastes a season.
A droplet is water at an activity of 1.0
Microbial growth tracks water activity (aw), the availability of unbound water, rather than total water content. ISO 29621 treats a cosmetic below 0.75 as microbiologically low risk, and the approximate biological floor for growth of any organism sits near 0.6. Those numbers are worked through, with the organism by organism minima, in do balms need preservatives, and the options once a product does need protecting are in preservatives for anhydrous products.
What those numbers cannot do is describe a contaminated balm, and this is the single most important idea on the page. Water activity is a bulk equilibrium property. A balm with a droplet of liquid water dispersed in it is not at equilibrium at all. Inside that droplet the water activity is 1.0, which is above the requirement of every organism that has ever been catalogued, and it stays 1.0 whatever a meter reports for the product as a whole. Averaging across a fat matrix and a puddle gives a figure that describes neither.
This is why a salve made with insufficiently dried herbs can grow visible mould at a plant fragment while the balm two centimetres away remains perfectly sound. The fragment is a wet island, and it carries its own microclimate and its own microbial inoculum. Low water activity elsewhere in the product prevents growth; it does not kill anything, and it does not reach inside a droplet.
Drying an infused oil properly
There are two defensible routes and they are not equally good. The better one is to remove the water before it ever meets the oil: dry the plant material until it snaps rather than bends, which for most leaf and flower material means an equilibrium moisture content of roughly 8 to 10 percent. Dry material cannot donate what it does not have, and it is the approach taken throughout the herbal salve formula.
The second route is to drive the water off after infusing, which is worth knowing because it also rescues an oil you have already made.
- Strain the oil thoroughly and let it stand. Twenty four hours at room temperature lets free water settle to the bottom, because water is denser than oil. Decant the clear oil off it. This alone removes the bulk of gross contamination without any heat at all.
- Heat the decanted oil to 100 to 110 C. Wide shallow pan, immersed probe, splatter guard. Steam has to leave the system, so nothing that condenses can be allowed to drip back.
- Hold until it goes silent, then two or three minutes longer. The endpoint is the absence of crackling, not the passage of a fixed time. A very wet oil can take twenty minutes; a nearly dry one goes quiet almost at once.
- Cool covered, in a dry room. A hot oil surface in a humid room reabsorbs moisture as it cools, which is a slow reversal of the work you just did.
- Add the antioxidant after cooling below 60 C. Tocopherol and rosemary extract do not survive the drying temperature, so they go in at the end, and the oxidative cost of the heat is discussed in rancidity and oxidation.
Be honest about what this costs. Holding an oil at 110 C is far below the smoke point of a refined oil, but it is not free: it accelerates oxidation, drives off the volatile aromatics that gave the infusion its character, and darkens some botanicals. Do it once, deliberately, rather than as a habit, and prefer dry raw material every time.
Wet tallow, and why rendering method matters
Rendered animal fat is the other classic carrier. Dry rendering, where the fat is melted on its own, leaves nothing behind but fat and cracklings. Wet rendering, where water or steam is used to lift the fat off the tissue, produces a fat that can hold a surprising amount of emulsified water even after it has been poured off, because proteins and phospholipids from the tissue act as emulsifiers. Rendered fat trade specifications control this with a combined moisture, insolubles and unsaponifiables figure, commonly capped around 1 percent, and the reason is that wet fat goes off.
Two practical tests. A tallow that spits when you warm a spoonful in a pan is not finished, whatever it looks like cold. And a cloudiness that clears when the fat is held liquid for an hour was water settling out; a cloudiness that persists in the liquid state is crystalline fat and is harmless. The composition and the rendering routes are covered on the tallow page, and the formulation consequences in tallow balm. If you render your own, finish every batch at 100 to 110 C until silent, strain hot, and pour into warm, bone dry jars.
Condensation, the source that arrives after you have finished
Every other source on this page is fixed by the time the lid goes on. Condensation is not, and it is the one that catches careful makers.
The mechanism is the dew point. Air at 20 C and 60 percent relative humidity has a dew point near 12 C, which means any surface in that room colder than about 12 C grows a film of liquid water within seconds. A jar taken from a 4 C fridge is far below that line. So is a tin that has been in an outbuilding overnight in winter. Once the film has formed on the balm surface, capping the container traps it inside with the product.
Two rules cover almost all of it. Never take a balm from cold storage into a warm room uncovered: leave it sealed until it has reached room temperature, which takes one to three hours for a small jar. And never cap a warm balm: cool the filled containers uncovered in a dry room until they are at room temperature throughout, then lid them. A tin capped at 45 C deposits its own condensate on the inside of the lid as it cools.
The same reasoning governs where a finished product lives. A bathroom shelf puts a cold container into warm, saturated air several times a day, which is the argument made in storing balms at home. Packaging choice matters too, because a jar that is opened with wet fingers gets both the water and the inoculum in one move, and a twist tube removes that route entirely. The trade-offs are in the packaging guide. One lookalike to rule out first: beads of clear liquid on a balm surface are often oil rather than water, which is syneresis and a different fault, diagnosed in balm sweating.
Can you dry the batch you already made?
Sometimes, and the conditions are strict. Driving water out of a finished balm means taking the whole batch above 100 C, which is well outside the normal 70 to 80 C melt window and hard on everything in the formula. It is justified only when all five of the following are true.
- The batch is less than about a week old. Long enough for growth to be unlikely, short enough that you know its history.
- No container has been opened, dipped into or sold. A finger in a pot adds an inoculum that heat will not make safe.
- Nothing smells wrong. Sour, musty, fermented or sharp all mean stop. Rancidity is a separate question, covered in rancid balm.
- There is no visible growth anywhere in the batch, including under the lids. One affected unit condemns every unit that shared the ingredient and the process.
- The formula contains no botanical fragments and no water bearing ingredient. You cannot dry honey or aloe out of a formula that lists them, and plant debris shelters organisms from the heat.
If all five hold, combine the batch, take it to 100 to 110 C with a splatter guard, hold until silent plus two minutes, cool to below 60 C, add the heat sensitive phase again because the original will have gone, then repour into dry, preheated containers. Expect to lose fragrance and some colour. If any of the five fails, discard it, and treat the infused oil, the rendered fat and the container stock as suspect until you know which of them carried the water.
Heating does not sterilise a balm. Low water activity stops organisms multiplying but does not kill them, and bacterial spores persist in a fat matrix more or less indefinitely, resuming growth whenever water returns. A dried batch is a dormant batch, not a clean one. That is why the five conditions above are about the batch's history rather than its temperature.
The rule, and what a bench cannot verify
The working rule is short: water is removed before the ingredient becomes a balm, never after. Dry the herb, dry the oil, dry the fat, dry the jug, cool uncovered and cap cold. Every one of those steps is cheap at the time and expensive to reverse.
The limits are worth stating plainly. A domestic bench can detect free water reliably at around 0.05 percent by heat and sound, and it cannot measure water activity at all without an instrument. No published water activity values exist for shea butter, cocoa butter, beeswax or candelilla wax, so anyone quoting a confident figure for a finished balm is extrapolating. If you sell, that gap is filled by measurement rather than argument: a benchtop water activity meter reading on your own worst case formula, and the stability and challenge testing described in shelf life testing. If you make for yourself, the honest position is that a properly dried anhydrous balm is very hard to spoil, that a wet one is easy to spoil, and that you can tell the two apart with a pan, a probe and your ears.
Frequently asked questions
Why is my balm cloudy or milky instead of clear?
Almost always water dispersed as fine droplets. Droplets around 0.4 to 0.7 micrometres across scatter visible light strongly and at all wavelengths equally, which reads as white. Confirm it by heating a small sample past 100 C: crackling and spitting mean water. Cloudiness that clears when the fat is held liquid is crystalline fat instead, and is harmless.
How do I know if there is water in my infused oil?
Heat 20 to 30 g in a shallow pan to about 110 C with a probe in it and listen from 95 C. Crackling or spitting is water flashing to steam, and the same principle underlies the official hot plate method for moisture in fats. Silence through 110 C means no free water, though it does not rule out a few hundred parts per million dissolved in the oil.
What temperature dries an infused oil?
Hold it at 100 to 110 C until the crackling stops, then two or three minutes longer. The endpoint is silence rather than a fixed time, so a wet oil may take twenty minutes and a nearly dry one almost none. Use a wide pan and a splatter guard, never a lid, because a lid condenses the steam and returns it to the oil.
Can water in a balm actually make it go mouldy?
Yes, and this is the reason it matters. A dispersed droplet has a water activity of 1.0 inside it regardless of what the bulk product reads, which is above the requirement of every known organism. Growth becomes possible from about 0.6 upwards, and most moulds need 0.80. A wet plant fragment in a salve is a self-contained habitat.
Is it safe to remelt a balm that got water in it?
Only if the batch is under a week old, unopened, unsold, free of any off smell and free of visible growth, and contains no botanical fragments or water bearing ingredients. Then take it above 100 C until it stops spitting. Heat does not sterilise a balm: low water activity prevents growth rather than killing organisms, so a dried batch is dormant, not clean.
Why does condensation form on my balm in the fridge?
Because a cold surface below the dew point of the surrounding air collects liquid water. Air at 20 C and 60 percent relative humidity has a dew point near 12 C, so anything straight out of a 4 C fridge is well below it. Let cold containers reach room temperature sealed, and never cap a balm while it is still warm.
Does a small amount of water always ruin a balm?
Not always, but it changes the product. Cosmetic damage such as haze or a dull set is often reversible by drying. What is not reversible is time: water plus a few warm weeks plus an inoculum gives growth, and no heat treatment makes that batch acceptable again. Judge by history and smell, not by appearance alone.
Sources and further reading
- International Organization for Standardization, ISO 29621, Cosmetics, Microbiology, Guidelines for the risk assessment and identification of microbiologically low-risk products, Geneva.
- International Organization for Standardization, ISO 8534, Animal and vegetable fats and oils, Determination of water content, Karl Fischer method, Geneva.
- American Oil Chemists' Society, Official Methods Ca 2c and Ca 2d, Moisture and Volatile Matter, hot plate and vacuum oven methods, Urbana.
- US Food and Drug Administration, Hazard Analysis and Risk-Based Preventive Controls for Human Food, draft guidance, Appendix 3, for minimum water activity values for growth.
- US Centers for Disease Control and Prevention, Botulism prevention, for home prepared oils infused with fresh garlic or herbs.
- European Directorate for the Quality of Medicines, European Pharmacopoeia general method 2.5.12, semi-micro determination of water, Strasbourg.
- Bohren, C. F. and Huffman, D. R., Absorption and Scattering of Light by Small Particles, Wiley, New York, for the size dependence of light scattering by droplets.
Reviewed and updated 6 September 2026. Spotted an error? Tell us and we will fix and log it.