My Soy Candle Looked Perfect After Pouring but Developed Tiny Oil Droplets on the Surface Three Days Later — Is the Wax Rejecting the Fragrance Oil?

What droplets on day three usually mean

When you pour, fragrance oil is dissolved in liquid wax. As the wax cools it sets into a solid, and in soy that setting commonly carries on for days: the surface you admired on day one is still settling on day three. If the finished solid can hold all the oil you put in, at the temperatures it meets, the surface stays dry. If it cannot, some oil ends up on the surface as droplets or a film. That is sweating, and it is a statement about this oil, at this load, in this wax, in these conditions, not about the wax alone.

That is why "rejecting" is the wrong frame. A wax does not decide to push an oil out on day three. In most cases the oil was already at or near the edge of what the set wax holds, and the day-three conditions (the wax finishing its set, a warm afternoon followed by a cooler night, an AC room switched on and off) tipped it over. Change any one of those and the same wax and oil can stay dry, which is exactly what the four-jar split exploits.

Three things in the story are worth writing down before you touch anything. Where on the surface: evenly across the top, in a ring around the wick, or at the glass edge. Which units: all of the batch, the last few poured, or only the jars that cooled in one place. What happened between pour and day three: where the tray sat, whether a fan or AC was on, whether the lids went on while the jars were warm. Each answer points at a different rung, and the location answers hand some cases straight to a sibling post: a ring round the wick is the wick-zone slice test, and oil under the base is the clear-jar 2×2.

Blot and look: is it oil at all?

Three different things look like "droplets" on a young soy candle, and only one of them is a formula problem. Before you change a load or a wax, identify what is on the surface. This is the step 0 box at the top of the ladder.

The sweat-cause ladder, rung by rung

Once the blot says oil, work down the ladder in the diagram. The order is editorial judgement about how common each cause is in makers' reports, most common first; CSI publishes no failure statistics. Here it also happens to be close to cheapest-to-check first, because the top rungs are read from your batch sheet and storage notes. The table sets out what rules each rung in and what rules it out.

Rung 1 — the load, and the basis slip behind it

CSI's convention is load as a percentage of total candle weight: a 150 g candle at 10% is 15 g of oil and 135 g of wax; at 8% it is 12 g and 138 g. The wax pages give a maximum but do not say which basis it is on. Read on wax weight, the conservative reading, CSI 464's 10% becomes 9.09% of total; our total-weight 10% is 11.11% of wax. A maker working at 10% of total is therefore at the edge on one reading and just past it on the other. The load-basis maths is in the load-basis hub. Past the basis there are the slips: oil measured in ml or by the cap, a small batch rounded up, or a "splash extra for evaporation". Each adds oil you did not plan. The full bracket that locates an oil's practical capacity is the sweat bracket; the 8-versus-9% tie-breaker is 3224.

Rung 2 — the conditions the candle met

Temperature cycling is the commonest trigger in makers' reports from Indian conditions: a warm afternoon and a cooler night, AC on for the evening and off in the day, a shelf that takes the sun for two hours. A formula near its edge can hold steady in a cupboard and sweat on a window shelf. If your droplets arrived after a hot spell, or only on units stored in one place, suspect this rung, and prove it with the heat-correlation log rather than by memory.

Rung 3 — the method on pour day

Oil added to wax that has cooled too far tends to incorporate poorly, and a short stir leaves it uneven. Neither shows on day one. Check your batch sheet: if it has no add temperature and no stir time, rung 3 is unknown, not ruled out. Follow the wax maker's figure: CSI 464's page states a pour of 80–90°C; in the 150 g jar, whose page says "Safe for hot pour up to 85°C", work at 80–85°C. CSI 468's page states a pour of 58–60°C. The visual checks for a cool add are the too-cool incorporation post.

One customer's review of the Mini Electric Wax/Soap Melter, written from a cold region, puts the problem in a line: "My temperature would drop within seconds". The pot temperature at the moment the oil goes in is a real variable, not a detail, and on a cold morning it can fall faster than you expect.

Rungs 4 to 6 — the oil, the cooling, the extras

Oils differ in how much of them a given wax holds; CSI does not publish oil composition, so this is observed, not predicted. If one scent sweats and others at the same load do not, the fair comparison is the same-load range map. Cooling conditions (a fan across the tray, an AC vent, lidding warm jars) are cell D of the split. And extras added on top of the fragrance, such as liquid dye or a second oil, change the mix; each has its own test in this cluster.

The four-jar split: one pour that names the rung

The split re-pours your formula four ways from the same wax bag and the same oil bottle as the sweating batch, so the wax and oil cannot differ between cells. Each cell changes one thing against its neighbour. Nothing is burned; you read surfaces. The worked example uses 10% (the reader's load at CSI 464's stated ceiling) and 8% as the step down, with Eucalyptus Spearmint's 100 g price standing in for your oil.

Why these four. A against B asks whether your method put more oil in, or less well, than you think. B against C asks whether this oil is simply over what the wax holds at 10%. B against D asks whether the room did it. If your batch sheet was already weighed, hot and stirred, A and B become duplicates, which is no loss: two jars agreeing is stronger than one. Pour all four in the same session, cool them side by side for the first hour, then move D to the original spot. Label the bases.

Costs are worked examples and exclude labour, freight, packaging and GST. If you own the wax, jars, wicks and oil from the sweating batch, the split needs no new cash at all; starting from nothing it is ₹1,864.40. The split jars are ordinary candles: if they stay dry and pass your burn tests later, nothing is wasted.

Reading it: day 1, 3, 7 and 14

Blot each jar on day 1, 3, 7 and 14 with a fresh labelled square, and write "clean" or "oily" plus the day the first oily blot appeared. Your original droplets came on day three, so day 3 is the first real reading and day 14 the verdict; the Eucalyptus Spearmint page's "Cure time is 48-72 hours for soy wax" is a burn-testing cure, not a sweat clock. For a finer surface grade across repeated hot-cool cycles, use the 0–3 score in the cycling protocol.

One jar per cell is a screening test. Before you change production, repeat the winning comparison with a duplicate: duplicates agree or you don't decide. If the answer is rung 1, the per-candle difference is real money in your favour: 8% instead of 10% saves 3 g of oil per 150 g candle, ₹9.20 of oil a candle at Eucalyptus Spearmint's 1 kg price and ₹920.40 per 100 candles, less the 3 g of extra wax (₹1.52 a candle in CSI 464).

What not to buy for day-three droplets

We sell everything in this box. We would still rather you ran the split first, because on rungs 1 to 3 the fix is less oil, a thermometer reading or a different shelf.

Four of the seven rows cost nothing. If you are tempted to fix sweat by pushing more wax into a sweating batch, the arithmetic and the restarted cure are in the remelt-or-reformulate post; if a full production lot is already sweating, grade it first with the 100-candle disposition procedure (3239 in the map below).

What to buy, if the split points to a purchase

Five items in ladder order: two tools for rungs 1 and 3, a stirring stick for a repeatable method, the wax the worked example uses, and the swap wax for rung 4. Chosen on what the product pages state, price per unit and stock on 26 September 2026; CSI has no measured sweat or throw data, so nothing is ranked on performance.

Map of this cluster: 3221–3240

This page owns the short answer, the blot and the ladder. Each of the other nineteen posts owns one test for one situation, so go straight to the one that matches what you saw.

The first-level version of this question, with the ₹ cost of the load you may be wasting, is the sweating first-look post; whether to change the oil if it keeps separating is the change-the-oil decision. If your candle also throws weakly, the neighbouring hub on strong cold, faint hot comes first.

Frequently asked questions