Balm troubleshooting

Why one formula gives you a different balm every time you make it

Volume measuring, scale resolution, long melts and wax lot variation all move firmness. The causes ranked, a batch record that finds yours, and the tolerances.

You made the same balm in March and in June, from the same written formula, and one sets firm enough for a tube while the other stays soft in the pot. The formula is not the variable, because it did not change. Something in the way the numbers became grams did, and there are only about six candidates. This page ranks them by how far each one can actually move finished firmness, then gives you the record that identifies yours in two batches rather than twenty.

Short answer

Rank the suspects: measuring by volume can be 65 percent out on wax, scale resolution costs 2 percent on a 50 g batch, melt and transfer losses 1 to 5 percent, wax lot variation 2 to 3 C of melting range. Weigh everything on a scale reading to 0.01 g, cap the melt at 30 minutes, and log lot numbers, melt peak, pour temperature and room temperature.

  • Beeswax pastilles: 0.59 g/mL bulk
  • 0.01 g scale under 5%
  • Melt cap 30 min
  • Reconcile yield to 2%
  • 10 day hold before sign-off

Rank the suspects before you change anything

The instinct is to adjust the formula. Resist it: a change made to compensate for an uncontrolled variable only moves the scatter to a different average. Find the variable first. These are the candidates, ordered by the shift each can produce in a typical 10 percent wax balm.

Sources of batch to batch variation, ranked by the size of the firmness shift each can cause. Percentages are of total batch weight unless stated. The firmness column translates the error into rungs on the ladder in balm too hard or too soft, where a rung is 3 to 5 percentage points of wax.
CauseTypical errorEffect on firmnessConfirm by
Measuring wax by volumeUp to 65% on the wax chargeTwo or more rungs, in either directionWeigh a tablespoon of your pastilles and compare with your assumed figure
Scale resolution and linearity1 to 2% of batchUp to a rung on a small batch, less on a large oneReweigh a known check weight at the bottom of the scale's range
Melt and transfer losses1 to 5% of batchHalf a rung, always in the soft directionReconcile poured yield against batch weight
Wax lot variation2 to 3 C of melting rangeRoughly the equivalent of 1 to 2 points of waxMake one batch from each lot side by side
Cooling rate and pour temperatureProcess, not compositionFeel and gloss change more than hardnessLog pour temperature and room temperature
How you judge the resultEntirely subjectiveNone, but it invents variation that is not thereUse a weighted penetration test at a fixed temperature

The top three are arithmetic and handling, not chemistry. Most makers start at the bottom of that table, adjusting pour temperatures and blaming suppliers, when the error is upstream in how the grams were arrived at.

Volume measuring is the single largest error available

A formula written in tablespoons and cups is not a formula, it is a rough intention. The reason is bulk density. Solid beeswax is 0.958 to 0.970 g/cm3 at 15 C, but pastilles are a packed bed with air between them, and their bulk density is about 0.59 g/mL. Convert a pastille volume using the solid density and you overestimate the mass by about 1.65 times.

The consequence is not subtle. An 11.6 percent salve worked out that way lands near 18 percent, as the worked conversions under batch scaling show: a firm stick rather than a salve, two rungs up the ladder from one arithmetic error. Worse, the error is not constant between batches, because it depends on pastille size, how the bed settles, and how firmly the spoon was pressed. Grated block wax, small pastilles and large pastilles do not pack alike, so the same tablespoon delivers different masses on different days.

Liquid oils behave better, but they are not 1.00 g/mL either. Most carrier oils sit near 0.91 to 0.92 g/mL, so a millilitre measure of oil is about 8 percent lighter than the same figure in grams, which on its own shifts a 10 percent wax formula to about 10.8 percent. The formula converter handles the arithmetic in both directions, but the correct answer is to stop converting and weigh.

Try this

Weigh one level tablespoon of your own wax, three times, and write the mean and the spread on the container. If the three weights disagree by more than a few percent, you have measured your own volume error directly, in your own workshop, with your own wax.

Scale resolution: the arithmetic on a 50 g batch

A kitchen scale reading in whole grams rounds every weighing to the nearest gram. On a 50 g batch that is 2 percent of the total batch weight from a single rounding, and the error compounds across every component you weigh.

Follow it through. A 50 g batch at 10 percent wax needs 5.0 g of wax, and on a 1 g scale anything from 4.5 to 5.4 g displays as 5 g. That is a wax charge between 9 and 10.8 percent, and the same rounding on the oil pushes the ratio further either way. Twelve percent against nine percent is a soft stick against a soft pot balm, which is the complaint that brought you here. On a 500 g batch the same resolution is 0.2 percent and vanishes into the noise, which is why the fault seems to disappear when you scale up.

Resolution is only half the story. Readability, the number a scale displays, is not accuracy. Inexpensive scales are least accurate at the bottom of their range, drift with temperature, and often specify an accuracy of two or three display increments. A 5 kg scale reading to 1 g is a poor instrument for a 50 g batch whatever its display suggests.

Scale readability needed to keep the rounding error on any single component below about 1 percent of that component's own mass. Derived from the rounding arithmetic, not from a standard.
Component massReadability neededWorst-case roundingTypical case
100 g and above1 g0.5%Oil phase of a kilogram batch
10 to 100 g0.1 g0.5%Wax and butter in a 250 g batch
1 to 10 g0.01 g0.5%Wax in a 50 g batch, most actives
Under 1 g0.001 g0.5%Flavour, tocopherol, pigment

The working rule that falls out of this: anything below 5 percent of the batch gets weighed on a scale that reads to 0.01 g, and anything below 1 percent of a small batch needs 0.001 g or a dilution weighed as a portion of a larger premix. Most makers need two scales rather than one, and both need checking against known weights on a schedule, which is what weighing and calibration covers. Buying a scale by its display resolution alone is the commonest equipment mistake in the workshop kit list.

What a long melt actually loses

The folklore is that oils evaporate during a long melt. Mostly they do not. Triglycerides and wax esters have negligible vapour pressure at 70 to 80 C, so a plain base held for an hour loses almost nothing to the air. Three other things do leave, and two of them matter.

  • Water. Beeswax and unrefined butters carry a few tenths of a percent of moisture, which drives off as the melt is held. Small, real, and one reason a long melt weighs less than what went into it.
  • Volatiles you added too early. Fragrance, essential oils, menthol and camphor leave readily at melt temperatures, and menthol and camphor sublime. Adding them at the start of a 45 minute melt rather than at 50 to 60 C at the end changes the strength of the finished product by an amount you cannot predict, which is a large part of why two batches smell different.
  • Material left on the vessel. This is the big one, and it is not evaporation at all. A film of solidified fat on the walls of the melting pot, on the stirrer, on the thermometer probe and in the pouring jug is routinely 1 to 5 percent of a small batch.

Wall residue is worse than a simple loss, because its composition is not the batch's. The film cools first, on the coolest surface, so the highest melting components crystallise into it preferentially. What stays behind is enriched in wax and hard butter, and what you pour is depleted in exactly the components that set the firmness. The effect always runs soft, and it is larger on a small batch because the surface to volume ratio is higher, which is the same geometry behind formulas that fail after scale up, running the other way.

Note

Cap the melt at 30 minutes at temperature, and hold it at 70 to 80 C rather than higher. That is long enough to erase crystal memory in most bases, short enough to limit residue and moisture loss, and below the point where beeswax darkens and its odour shifts. The full window and what each end of it costs you is under pour temperatures.

The check is a yield reconciliation. Weigh the empty vessel, weigh it with the finished melt, weigh it again after pouring, and compare what you poured against what you weighed in. Anything worse than 2 percent unaccounted for is worth chasing, and the same figure is what batch records ask you to log. Weighing components directly into the melting vessel rather than into intermediate cups removes a whole class of these losses in one change.

Wax lot variation, and what 2 to 3 C actually does

Beeswax is a natural material and the pharmacopoeial monographs specify a band, not a point. USP gives a 62 to 65 C melting range for yellow and white wax, and the European Pharmacopoeia monographs for Cera flava and Cera alba specify a drop point range in the same region. Two compliant lots can therefore sit 2 to 3 C apart, and differ in composition too: the ratio of hydrocarbons to monoesters and the free fatty acid content vary with colony, forage, season and rendering.

In a formula carrying 10 percent wax, that difference is not visible as a melting point on the finished balm, it shows up in when the structuring network starts to form during the cool and how dense it ends up. A higher melting lot begins to crystallise earlier in the cooling curve and sets a little firmer. The magnitude is an estimate rather than a measurement: expect something near one to two percentage points of wax equivalent between two compliant lots. The direction is reliable, the size is not published.

The management is straightforward. Buy wax in larger lots so a formula runs on one lot for as long as possible, record the lot number on every batch, and when a lot changes, run one batch from the old and one from the new in the same session before committing production. That comparison converts an argument into a measurement, and it is the part of sourcing ingredients most makers skip. The same applies to butters, where shea grain is lot sensitive, and to candelilla and carnauba, which tolerate less substitution than beeswax does.

The diagnostic: make it twice in one session

This is the fastest way to split the problem in half, and it takes one afternoon.

  1. Make the same formula twice, back to back, same session. Same lots, same vessel, same thermometer, same room, same cooling location. Weigh everything and write down the weighed figure rather than the target.
  2. Pour both into identical containers and set them side by side, not one on the worktop and one on a cold windowsill.
  3. Leave both 24 hours at a recorded room temperature, then measure firmness with the same method on both, at the same temperature.
  4. Read the result. If the two agree closely, your process is repeatable and the variation lives between sessions: lots, ambient temperature, or a technique that has drifted. If the two disagree, the variable is inside a single session: weighing, transfer losses, or how consistently you cool.
  5. Then change one thing. Repeat the pair with the suspected variable altered and nothing else. Two variables changed at once tells you nothing.

Disagreement within a session points at the mechanical causes in the first half of this page. Pairs that match within a session but differ between sessions point at lots and ambient conditions, which is a different fix: buy in bigger lots and control the room, rather than buy a better scale.

Measure firmness the same way every time

Half the reported inconsistency is inconsistent judging. A finger press at 19 C on a Tuesday and at 24 C on a Saturday reports a difference the balm does not have, because firmness is strongly temperature dependent across normal room temperatures.

A weighted penetration test removes that. The principle is the one behind the standard needle penetration methods for waxes, such as ASTM D1321: load a probe with a fixed mass, let it sit for a fixed time at a controlled temperature, and record how far it sank. A bench version uses the round handle end of a teaspoon or a blunt rod of known diameter, a small weight taped to it, five seconds of dwell, and a depth read off a ruler.

The absolute number will not match anyone else's rig, and does not need to. Fix the probe, the mass, the dwell time and the sample temperature, record all four alongside the reading, and the number becomes comparable between your own batches. Instrumented versions are under measuring balm hardness. If what you are chasing is feel rather than firmness, a structured panel beats a penetrometer, and sensory testing sets out how to run one.

The batch record that finds your variable

None of the above works from memory. Six fields, recorded every time, identify almost any consistency problem within four or five batches, because they let you sort batches by each variable and look for the pattern.

The minimum fields for diagnosing inconsistency, over and above the formula itself. These are a subset of the full record set in batch records and traceability, chosen for this problem.
FieldRecordWhat it explains
Lot numbersEvery raw material, including the waxSupplier variation, and it is your only route back if a lot turns out to be the cause
Melt peakHighest temperature reached, from a probe in the fatWhether crystal memory was erased, and whether you overheated
Hold timeMinutes at temperature before pouringResidue and volatile losses; the reason a batch runs soft
Pour temperatureReading at the moment of pouringGloss, shrinkage, tunnelling, layer separation in tubes
Room temperatureAmbient at pour and at setCooling rate, which sets the crystal population
Set timeMinutes to a firm surface, and where the units satConfirms the cooling rate you assumed actually happened

Record the weighed figures rather than the targets, and add the poured yield. "5 g wax" because that is what the formula said is worth nothing; "5.02 g wax, lot 24B, yield 47.1 g of 50 g" is a diagnostic instrument. The batch calculator generates the target column, leaving you only what actually happened to fill in.

Careful

Temperature records are only as good as the probe. Two thermometers disagreeing by 5 C will make a well controlled process look erratic on paper and send you chasing a variable that does not exist. Check yours at the ice point before you trust any of the temperature fields above; thermometer readings that disagree covers the calibration and the immersion depth error that causes most of it.

Tolerances worth holding, and the ones that are not worth chasing

Consistency is not free, and some of it is not worth buying. These are the tolerances that pay for themselves in a small workshop.

Practical tolerances for a small balm workshop, with the failure each one prevents.
VariableHold toWhy that number
Structuring phase, wax and hard butterPlus or minus 1% relativeBelow the level that moves a rung on the firmness ladder
Minor components under 5% of batchPlus or minus 2% relativeAchievable on a 0.01 g scale, and they carry less structure
Melt peak temperaturePlus or minus 3 CWide enough to be achievable, tight enough to erase crystal memory reliably
Hold at temperature30 minutes maximumLimits residue, moisture and volatile losses
Pour temperaturePlus or minus 3 CThe band within which gloss and shrinkage stay stable
Yield reconciliationWithin 2%Anything worse means losses large enough to shift composition
Room temperature at setPlus or minus 4 CRealistic for a domestic workshop; log it rather than fight it

Not worth chasing: the last half degree of pour temperature, or the exact stirring speed. One caution belongs to hygiene rather than precision. A vessel still holding residue from a heavily fragranced batch contaminates the next one, and the fix is a cleaning routine rather than a tighter tolerance, as covered under workshop hygiene.

Hold the batch ten days before you sign it off

The last cause of apparent inconsistency is judging too early. Fat crystal networks are not finished when a balm feels set: transformation and coarsening continue for days, so a balm poured on Monday can be measurably firmer, and sometimes grainier, by the weekend. Two batches assessed at different ages disagree even when identical.

Sign off on a fixed schedule instead. Assess at 24 hours as an early warning, then hold ten days at a recorded room temperature and assess again before releasing the batch or judging the formula. Keep two retained units so a later question can be answered against the material rather than a memory. The reasoning behind the ten day figure is under crystallisation and cooling rate, and the retain regime under shelf life and stability testing.

The decision rule, and what you cannot control

Work the list top down and stop when the scatter closes: weigh instead of measuring, match scale resolution to your smallest component, weigh into the melting vessel and reconcile the yield, cap the melt, log the six fields, compare pairs made in one session. Only after all six is the supplier a reasonable suspect, and by then you will have the records to demonstrate it.

Two honest limits. Some variation is real and cannot be removed by a small maker: natural wax and unrefined butter differ between lots, and a workshop without climate control cools a batch differently in January and July. Widen your specification to something you can genuinely hold rather than claim a precision you do not have. And a formula sitting at the edge of a band amplifies everything above it. If 10 percent wax is a soft pot balm and 12 percent is a stick, a formula written at 11.5 percent will look inconsistent however well you weigh. Move the target to the middle of the band you want, using the ladder in balm too hard or too soft, and much of the apparent inconsistency stops being visible.

Frequently asked questions

Why is my lip balm different every time I make it?

Almost always the way the grams were arrived at rather than the formula. Measuring wax by volume can be 65 percent out because pastilles pack with air between them. A whole gram scale on a 50 g batch rounds by 2 percent of the batch. Material left on the melting vessel runs another 1 to 5 percent, always in the soft direction.

What scale do I need for making balm?

One that reads to 0.01 g for anything under 5 percent of the batch, and a larger capacity scale reading to 0.1 or 1 g for the bulk phase. Match resolution to the smallest thing you weigh, aiming for a rounding error under about 1 percent of that component. Readability is not accuracy, so check both against known weights regularly.

Does beeswax really vary between batches?

Yes. It is a natural material and the pharmacopoeial monographs specify a range rather than a point, with USP giving 62 to 65 C for the melting range of yellow and white wax. Two compliant lots can sit 2 to 3 C apart and differ in hydrocarbon to ester ratio, which shows up as slightly different firmness at the same use rate.

How long should I melt balm ingredients for?

Long enough to melt fully and erase crystal memory, and no longer. Thirty minutes at 70 to 80 C is a sensible cap. Longer holds increase moisture loss, drive off any volatiles already added, build up more solidified residue on the vessel walls, and start to darken beeswax and shift its odour.

How can I tell whether the problem is my process or my supplier?

Make the same formula twice in one session with the same lots, vessel and cooling position, then compare them. If the two match, your process is repeatable and the variation comes from between sessions: lots, ambient temperature, or drifted technique. If the pair disagrees, the variable is inside your process, most often weighing or transfer losses.

How soon after pouring can I judge a batch?

Not on the day. Fat crystal networks keep developing for days after a balm feels set, so a batch assessed at 24 hours and one assessed at two weeks will disagree even if they are identical. Check at 24 hours as an early warning, then hold ten days at a recorded room temperature and make the real assessment then.

Is a 2 percent weighing error really enough to notice?

On a small batch, yes, because the error lands disproportionately on the structuring phase. Two percent of a 50 g batch is 1 g, and on a 5 g wax charge that is a fifth of the wax, taking a 10 percent formula to about 12 percent. That is the difference between a balm that scoops from a pot and one firm enough for a tube.

Sources and further reading

  1. United States Pharmacopeia, USP-NF monographs: Yellow Wax and White Wax, melting range and identity requirements.
  2. European Directorate for the Quality of Medicines, European Pharmacopoeia monographs: Cera flava and Cera alba, Strasbourg.
  3. Bogdanov, S., Beeswax: production, properties, composition and control, Bee Product Science, on compositional variation between rendered wax lots.
  4. International Organization of Legal Metrology, OIML R 76-1, Non-automatic weighing instruments: metrological and technical requirements, on verification scale interval and accuracy classes.
  5. ASTM International, ASTM D1321, Standard Test Method for Needle Penetration of Petroleum Waxes, West Conshohocken PA.
  6. International Organization for Standardization, ISO 22716:2007, Cosmetics, good manufacturing practices, on batch documentation, weighing and yield reconciliation.

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