Swirls, streaks and patches of colour mean the pigment never properly dispersed

Streaks are pigment agglomerates plus a melt that set before it was even. Grinding into castor, a drawdown check and the pour temperature that keeps colour flat.

A tinted balm that left the jug as one colour and set as two or three has a mechanical problem, not a formulation one. Either the pigment arrived in the melt as clumps that stirring was never going to break, or it was properly dispersed and then frozen into a pattern by a melt that began setting before the last container was filled. The two faults look similar in the finished tin and need opposite corrections, so the first job is telling them apart.

• Cause: agglomerates or a setting melt
• Paste: 1 part pigment to 3-5 castor oil
• Fill above 65 C
• Remelt safe for mineral pigments

Three faults wearing the same coat

"Streaky" covers three different faults: a dispersion fault, a thermal fault, and a crystallisation fault that has nothing to do with colour at all. Reprocessing helps two of the three, so read the pattern before you melt anything.

Discrete specks in an otherwise even field are undispersed pigment. Broad swirls with a soft boundary, like poorly stirred paint, are a melt that was partway through setting while you were still moving it. Cloudy patches with no colour boundary at all are usually fat crystals rather than pigment, and belong with balm sweating and the cooling behaviour in crystallisation and cooling rate.

Agglomerates you can break, aggregates you cannot

Pigment powder is not a bag of individual particles. Iron oxide and titanium dioxide primaries are in the range of a fraction of a micron, and particles that small have enormous surface area relative to their mass, so they stick to each other the moment they are made. They stick in two different ways, and the difference decides what grinding can achieve.

An agglomerate is a loose assembly of primaries held together at edges and corners by van der Waals attraction, with air trapped in the voids between them. Its total surface area is roughly the sum of the surfaces of the particles in it, which is another way of saying the particles are touching rather than joined. Shear breaks agglomerates apart, and that is the entire purpose of grinding a pigment paste.

An aggregate is different. During manufacture, particularly the calcination that gives iron oxides their final shade, primaries fuse face to face and the assembly behaves as one particle with far less surface area than its parts. Nothing you can do on a glass tile will separate it. That is why pigment grade matters more than grinding time past a point: a coarse, heavily aggregated iron oxide has a floor on how fine it will ever go, and that floor is set in the pigment plant. If a paste refuses to come smooth after ten minutes of real work, the powder is the problem, not the technique.

Dispersion is conventionally described as three stages, and the craft habit of tipping powder into a hot melt attempts only the last one. First the liquid has to wet the powder, displacing the air that is holding the agglomerates together. Then mechanical shear has to separate what wetting has loosened. Then the vehicle has to stabilise the result so the particles do not simply find each other again. Stirring a melt provides a trickle of the second stage and none of the first.

Wetting is the step that stirring cannot do

Molten balm at 70 C is thin. Vegetable oils at that temperature sit around ten to twenty millipascal seconds. Castor oil at room temperature sits around six hundred to nine hundred, and that gap is the mechanism.

Shear stress is viscosity multiplied by shear rate. A spoon turning in a thin hot liquid manages a shear rate of a few tens per second, and at ten millipascal seconds the stress that produces will not pull an agglomerate apart. A spatula worked against a glass tile, in a paste two orders of magnitude more viscous, transmits real force into the clumps instead of flowing around them. That is why the paste has to be thick and why it has to be made cold: heating castor oil to thin it defeats the only advantage it has.

Castor oil also wets mineral pigment better than any other common balm oil, because ricinoleic acid carries a hydroxyl group on the chain. That makes the oil polar in a way that ordinary triglycerides are not, so it has an affinity for the polar oxide surface of the pigment and can displace air from it. The full behaviour of the oil is covered on castor oil, and the approval side of pigment choice, which comes before any of this, is on colourants and tints.

1. Weigh pigment and castor oil at 1 part to 3 to 5 parts. That is a 17 to 25% paste, thinner than the 25 to 40% masterbatch used for a lipstick, because a lip balm carries less pigment and a thinner paste is easier to work by hand without a muller.
2. Wet the powder before you grind it. Fold the oil in until there is no dry powder visible, then leave it two minutes. Wetting is a slow process and you are letting it happen instead of fighting it.
3. Work it against a hard flat surface. A glass tile and a stiff palette knife, pressed down and drawn across, not stirred in a bowl. Five minutes of real pressure. A pestle and mortar works, a whisk does not.
4. Scrape and repeat. Gather the paste back into the middle every minute or so, because the material at the edge of the tile is not being worked.
5. Draw it down and look. The check in the next section. If it fails, another two minutes usually settles it.

The drawdown: a 60 second check before you commit a batch

The paint industry has judged dispersion this way for a century, and the domestic version costs nothing. Put a small blob of paste at one end of a piece of smooth white card, set the edge of a palette knife or an old credit card against it at a shallow angle, and pull it down the card in one steady stroke to leave a wedge of colour that runs from thick to nearly transparent. Then look at the thin end under good light, at a raking angle.

A finished paste draws down as a continuous film that fades smoothly to nothing. An unfinished one shows the agglomerates plainly at the thin end, as specks, as short scratch-like streaks trailing in the direction of the stroke, or as a grainy texture in what should be a flat wash. The thin end is where the test does its work, because the film there is thinner than the clumps you are looking for, so anything oversized stands proud and catches the light.

The industrial version is a Hegman gauge, a steel block with a groove tapering from around 100 microns to zero, read at the point where the coarsest surviving particles start to scratch. It turns a subjective look into a number you can write in a batch record. If you make tinted balm twice a year, white card is enough.

The cold paste problem, in numbers

Here is the fault that catches people who have done the grinding properly. The paste is perfect, it goes into the melt at room temperature as a cold slug of viscous oil, and the balm sets streaky anyway. Run the heat balance and the bulk numbers look reassuring, which is the trap.

A two degree bulk drop sounds harmless. It is not what happens. Mixing is not instantaneous, and for the first several seconds the paste is a 20 C blob with a 70 C liquid wrapped around it. Everything in the boundary layer passes through every temperature between the two, including the range in which the wax stops being dissolved and starts being solid. Beeswax congeals in the low sixties, so the shell of balm around a cold paste crystallises, traps pigment in whatever pattern it was in at that instant, and then gets stirred through the rest of the batch as soft solid lumps that never redissolve. That is a swirl, and no amount of further stirring removes it, because the material is no longer liquid.

Note the last two rows. Scaling changes the bulk arithmetic not at all but makes the local problem worse, because a 25 g slug of cold paste has a smaller surface to volume ratio than a 5 g one and takes longer to equalise. That is one of the quiet reasons a tinted formula that worked at 100 g goes wrong at 500 g, alongside the effects in batch scaling. The fix is trivial: stand the pot of paste in the same water bath while you weigh everything else, and the third row of the table becomes your situation instead of the second.

Over milling mica destroys the thing you paid for

Everything above applies to absorption pigments: iron oxides, titanium dioxide, carmine, lakes. Pearlescent mica is a different material and grinding it is actively harmful.

A pearl pigment is a thin flat platelet of mica coated with a controlled film of titanium dioxide, usually between about 60 and 160 nanometres thick. The colour is thin film interference from that coating and the sparkle is specular reflection from the flat face, so both depend on the platelet staying flat and staying large. Grinding snaps platelets into fragments, which cuts the reflecting area so the sparkle dies, exposes broken white edges so the mass tone goes chalky, and can damage the coating so the interference colour shifts. The result is a dull grey-beige powder.

It does not need grinding anyway. Mica is manufactured as a dispersible platelet and wets into oil readily. The only preparation it benefits from is a slurry in a little liquid oil so it does not float as dry powder on the melt. If a shimmer balm looks patchy rather than dull, the platelets floated or settled rather than failing to disperse, which is the physics of pigment sinking to the bottom.

Rescuing the batch in front of you

Remelting is safe for mineral pigments. Iron oxides and titanium dioxide are already fully oxidised inorganic solids with decomposition temperatures many hundreds of degrees above anything a balm sees, so nothing in the colour is harmed by a second pass through 70 C. The materials that suffer are the ones that always suffer: flavour, essential oils, tocopherol and any organic lake or carmine, which are less thermally robust than the minerals. Plan to re-dose the volatiles rather than hoping they survived.

1. Decide what you are fixing. You cannot get pigment back out of a balm, so if the fault is specks, remelt the batch and grind a fresh paste of the same pigment in a little extra castor oil, accepting a slightly stronger shade and a slightly higher castor content. If the fault is swirls from a thermal clash, the pigment is fine and you need only heat and stirring.
2. Scrape the whole lot into one vessel. Including the thin films left in tins. Rescuing container by container guarantees six different shades.
3. Melt to 70 to 75 C in a water bath and hold ten minutes. Long enough for every wax crystal from the failed pour to be genuinely gone, not merely invisible. Use an immersed probe, not an infrared reading off the surface, for the reasons set out in thermometer readings disagree.
4. Add any new paste at 70 C, pre-warmed. Paste into melt, never melt into paste, and stream it in while stirring rather than dropping it in as a lump.
5. Stir three minutes. Slowly, scraping the sides and base of the jug where paste concentrates. It feels absurdly long and is roughly what it takes.
6. Warm the containers. Ten minutes in a 40 C oven or on a warm tray removes the wall halo entirely.
7. Fill everything above 65 C, stirring between pours. If the batch drops toward the bottom of the window before you finish, put it back in the bath and bring it up again. The windows are on pour temperatures.
8. Cool undisturbed and judge in 24 hours. A warm balm looks darker and more even than a set one, so colour judgements made over the jug are always wrong.

Making the next batch flat by design

Four changes remove almost all of this permanently.

Keep a standing colour concentrate. Grind a large jar of paste at a fixed ratio, label it with the pigment, the lot number and the percentage, and dose from it by weight. This is the single biggest improvement available: it makes pigment loading reproducible to two decimal places on a decent scale, and it means dispersion quality is decided once, carefully, rather than at the end of a long evening. Castor oil is stable enough that a concentrate keeps as long as the oil does.

Sieve the concentrate. Push the finished paste through a 150 micron mesh, a number 100 test sieve in the US series. It breaks nothing, but it removes coarse aggregates and stray fibre, and the residue on the mesh is a running check on your pigment lot: one that suddenly leaves twice as much behind has changed.

Formulate for suspension, not just for colour. Pigment particles are several times denser than the oil around them and will settle in a thin melt. Castor oil at 5 to 15% of the formula raises the viscosity of the liquid phase and slows that down, and the wax network described in balm texture science locks everything in place once it forms. A formula that is 40% low viscosity ester and 8% wax will drop its pigment during a slow fill no matter how well you ground it.

Stir right up to the pour, and pour fast. Not vigorously, which whips in air and produces the fault covered in air bubbles in balm, but continuously. The interval between stopping the stir and the balm setting is the window in which pigment settles and swirls freeze, and the discipline of a fast fill, described in filling lip balm tubes, is worth more than any of the equipment sold for the job.

What a rescue cannot fix

Three limits worth being straight about. A badly aggregated pigment lot cannot be ground smooth, so if a drawdown still fails after ten minutes of proper work, change the pigment rather than the method. Broken mica platelets are finished, and the only route back is fresh mica. And a balm through two full remelts has had its volatiles driven off twice and its unsaturated oils heated twice, so however good the colour looks it is no longer the product you costed.

The decision rule is simple enough to apply while the jug is still warm. If the pattern is specks, the pigment was never dispersed and a remelt alone will not help. If the pattern is swirls or a wall halo, the dispersion was fine and the temperature was not, so a remelt with a hot paste and a fast fill will fix it. If the pattern is a dull cast rather than a pattern at all, you have ground the mica and the batch is a lesson rather than a product. Everything else on a tinted balm, from shade drift on the shelf to weak payoff, is a different fault with a different cause, and the ones most often mistaken for this are on tinted balm colour fading and drag and poor glide.

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