My First Candles From the Pouring Pot Are Fine but the Last Candles Show Oil Separation — Could the Batch Be Becoming Non-Uniform During Filling?

The pattern is the clue

This is one of the most useful patterns a maker can see, because it rules so much out. The first candles and the last candles came from the same wax, the same oil, the same load and the same stir. If the formula could not hold the oil, the first jars would weep as well. They do not. Something changed between the first pour and the last.

That something is the batch itself, during filling. A pot that was evenly mixed when you put the stirrer down is not frozen in that state. For the next however-many minutes it takes to fill every jar, it is cooling, sitting and being drawn from in a particular order. Each of those can make the last candles different from the first.

How a batch drifts while you fill

It cools

Unless the melter holds it steadily, the batch loses heat throughout filling, fastest at the walls and the surface. The first jars are poured at close to your intended pour temperature. The last may be poured noticeably cooler, from wax that has started to thicken. Cooler, thicker wax going into a jar can set differently, and oil that was comfortably held in hotter wax may have a harder time staying evenly distributed in the last few pours.

It sits

Between pitcher refills nobody stirs the pot. If the oil was not perfectly mixed to begin with, standing time gives it a chance to drift. Even if it was, the cooler wall wax does not stay mixed with the warmer centre on its own.

You draw from different places

Early refills tend to take wax from the top and centre. By the end, you are tipping the pot, pulling wax from the base and walls, and often scraping. That is where cooler wax, thin set skins and any oil that was never fully mixed collect. The last two or three jars are effectively a sample of the parts of the pot your stirring reached least.

The jars change too

Late jars are not always the same as early ones either. If the jars were warmed before filling, the last ones may have cooled back towards room temperature by the time you reach them. If they were stacked near a window, an air-conditioning vent or a cold floor, the end of the row may sit in a different spot from the start. A cooler jar chills the wax at the glass faster, which changes how the candle sets. Note where each numbered jar stood while it waited and while it cooled.

Log the fill, not just the batch

Next batch, number every jar on its base in pour order. At each pitcher refill, note the time, take a probe reading of the pot, and note whether you re-stirred. Then observe all the jars on the same suggested schedule, days 1, 2, 3 and 7.

The worked example: a 30-candle batch of CSI 464 with Lavender at 8% of total weight, 200 g candles. That is 30 × 200 = 6,000 g in total, with 6,000 × 0.08 = 480 g of oil and 5,520 g of wax. With the large 600 ml pitcher holding roughly 540 g of melted soy wax, about two candles per refill, the fill takes around fifteen refills. That is fifteen chances for the pot to change, and fifteen chances to correct it.

Fill every refill line, not only the five shown. Then look at where the trouble starts. In the example pattern from the diagram, jars 25 to 30 show oil. If your log shows the pot reading falling steadily and no re-stir after the main stir, you have found it. If the readings stay steady and you re-stirred throughout, and the late jars still weep, look harder at the dregs: what went into jars 29 and 30?

Filling so the last jar matches the first

Holding in the melter

If you fill from a melter, its holding setting matters. The CSI electric melter page states a 60–100°C display with a ±10°C deviation, so a steady display does not guarantee a steady batch. Leave the probe readings in charge. If the melter is switched off after the main stir to save power or to cool towards pour temperature, expect the batch to keep falling during the fill and plan re-warms rather than hoping the last jars arrive in time.

Melter shape also changes where the dregs come from. As the level drops, a tall vessel exposes more cooled wall above the wax, and the last litres sit in a base that has been losing heat for the whole fill. Stirring the base gently before the last few refills does more good than stirring at the start alone.

Oil gathering near the wick in late jars

Sometimes the late-jar fault shows as a small pool around the wick rather than droplets across the top. The wick area is often the last to set, so a slightly oil-rich, cooler pour can collect there. Treat it the same way: log it by jar number, and check whether it tracks pour order before calling it a separate problem.

What the late jars cost

The series' standard example prices this candle: 184 g of CSI 464 at ₹0.5074 per gram (₹93.36), plus 16 g of Lavender at ₹5.221 per gram (₹83.54), plus a matte jar at ₹70.80 and an Eco C1 wick at ₹5.90, giving 184 × ₹0.5074 + 16 × ₹5.221 + ₹70.80 + ₹5.90 = ₹253.60 per candle, excluding packaging, labels, labour, shipping and GST.

If the last six of thirty fail, that is 6 × ₹253.60 = ₹1,521.60 in materials lost. Spread across the 24 good candles, the batch costs 30 × ₹253.60 = ₹7,608.00, and ₹7,608.00 ÷ 24 = ₹317.00 per good candle instead of ₹253.60. A few seconds of re-stirring per refill is a very cheap way to recover that difference.

What to do with this batch

Separate the late jars and treat them as rejects or rework candidates. If the cause was cooling and drift rather than the formula, remelting them into the start of your next batch at the same load, properly stirred, is reasonable, with new wicks and cleaned jars, and the remelted candles observed again like any new batch. Do not add oil to them on a guess.

Hold the early jars too, and release them only when they have passed your full observation schedule. They are probably fine. Probably is not a dispatch standard.

Where this page fits

CSI already has guides that cover parts of this question. This page answers one sweating decision; these cover the neighbouring questions:

Three things to buy for this fix

Frequently asked questions